Today a common data center looks like a chaotic assembly of systems purchased at different times, and as a rule each separate server carries out a single specialized business task using only a small portion of its performance resource.

According to different estimations, an average server load comprises not more than 30%. However, despite of only small portion of resources being used, the hardware still requires the full time care and support, which results in unreasonable spending of budget on data center maintenance. This is for a reason that every piece of hardware requires individual attention.

This problem was designated a long time ago, and even some solutions had been developed. But only now we can state that there exist technologies, solutions and opportunities for implementing those ideas in practice for a scale of the entire data center, as opposed to implementation on the level of separate elements.

Data center consolidation concept
As we introduced the term data center consolidation, we can speak about a large united “organism”, which incorporates peer hardware and software resources, constantly controls execution of business applications and reallocation of computation power depending on requirements (See also "data center virtualization" term ).

In order to build such well-organized mechanism, both classical standardization approach and newest virtualization technologies should be used.

Standardization
Standardization is beneficial in regard to ownership value owing to minimization and unification of maintenance processes as well as introducing interchangeability of physical components.

Virtualization
Virtualization allows flexible redistribution of used resources among applications, providing for balance load, while introducing a centralized control mechanism for carrying out tasks such as updating and backup.

Based on practical experience, the most demanded mechanism of server standardization is purchasing blade systems, which allow enlarging the total computation resource of the data center to the needed level. Besides, IT virtualization tools currently offered at the market feature flexibility in balancing the load between these “bricks” of the data center.

When using data center consolidation, you must clearly know all items of expenses and the affecting factors in order to estimate IT costs. The first step in assessing these costs is listing the inventory of IT resources.

The total cost is comprised of all IT-related expenses summed up:

• expenses spent on hardware and software purchases,
• systems maintenance,
• administration and engineering costs,
• end user related expenses,
• losses due to failures and other.

While planning IT budget, both maintenance expenses and IT development costs should be considered.

With hardware and software costs being clear enough, the costs of other components (services, for example) are controversial. You should know not only the cost of the service, but also have the means to control its quality.

These tasks can be carried out by implementing IT Infrastructure Library (ITIL). This approach gives an opportunity to estimate costs of IT services, to control their quality and possibly switch to another provider thus controlling the service cost.

Company’s IT department turns from being a completely unprofitable subdivision to a business department with a well-organized system of internal mutual settlement of expenses and costs control.

It’s important to understand that if you save money today by refusing to create a scalable solution, then in future you will face the need of additional investments for business extension.

In case the service cost offered by third party provider is lower than the cost of own IT department, the service can well be outsourced.
Benefits of outsourcing include optimization of company’s expenses spent on IT department and own infrastructure, since when a provider’s infrastructure is used there is no need to maintain own one.

Therefore, company saves the money that otherwise would be spent on hardware, electricity, rooms, software and support.

Also, virtualization solutions allow cutting costs of electricity, improving controllability of the system, reducing recovery time, and in many cases reducing software licensing expenses.

VDR (virtual data room) – is a website created to serve a certain purpose for a specified period of time. It grants authorized users access to the protected document storage according to users’ rights.

Initially, virtual data rooms were intended for use as alternative to physical data rooms upon carrying out due diligence procedure upon mergers and take-overs.

It looked like this: the seller dedicated or rented one or more rooms, where all paper folders were to be gathered. An appropriate schedule was made for potential buyers, who could visit the rooms and work with the documents. But given a considerable amount of paper folders, the time needed to familiarize with documents was too much and substantially delayed the signing of agreements.

For what purpose and how a virtual data room can be used?
It can serve well in practically any situation that demands granting access to confidential information for several persons simultaneously, such as:

• corporate reports analysis;
• creating document archive;
• performing audit;
• preparation of initial public offering (IPO);
• mergers and take-overs;
• seeking and presenting information for investors;
• bankruptcy and restructuration;
• preparation to request certificates for medicines or other scientific products.

Alternatives
In our case these are traditional data room, FTP server or intranet portal.

Comparison to traditional data room

Virtual data room is a website, which solves the problem of concurrent access to documents for users having corresponding rights, and gives an opportunity to work with needed documents anytime and anywhere.
Comparing rental and maintenance expenses when using physical rooms, as well as time consumption for workers moving to and from the room can be easily imagined, and there are also expenses for security measures (like data encryption and personalized access). In addition to getting rid of all these expenses, virtual data rooms bring convenience of operation with automated directories, reports and search engines.

Comparison against FTP protocol
FTP protocol involves information transfer in open form, and thus it can be potentially retrieved by someone else between FTP server and the addressee. This protocol also introduces complications with controlling the accounts.

In general, FTP just gives a storage “place”, where “something” is stored, and no options like version tracking or chatting are provided.

Comparison against other tools

A similar functionality can be implemented on a corporate intranet portal; however, creating an intranet site from scratch takes much more time and money than opening a virtual data room. Besides, opening access to intranet for strangers can conflict with company’s security policy.

Of course, e-mail, fax, CDs and flash drives are out there. But only virtual data room can provide the following features: no limits for file size, optional encryption, version tracking, and feedback associated with certain documents. But the major advantage is the “magical formula” 24/7 with the opportunity to access the information stored in virtual room from any location on the planet connected to the Internet.

How to choose a provider?

At the first glance, choosing a virtual data room provider is just like choosing an internet provider, and therefore you can consider the following criteria:

• Does provider have any experience in implementation of projects similar to the one you need?
• Cost of service. It’s worth seeking through offers from several companies at the market, which provide this service.
• Availability of multilanguage interface is an important advantage for foreign companies.
• Availability of 24/7 support service, which speaks the language that you and your coworkers understand.
• Opportunity to purchase both retail “box” product and a customizable solution to accommodate your particular needs.
• Simplicity and usability of virtual data room interface will reduce the time spent on getting familiar with the service and minimize the count of mistakes.
• Whether provider is giving an option to convert information from paper to digital form.
• Degree of information protection.
• Option to select physical data storage location. It should be up to the customer where he decides to store the confidential data – on provider’s side or on your own servers.

Raid 5 data recovery is one of the most demanded services for the reason that level 5 RAID-arrays are keeping the leading positions in terms of popularity among owners of servers and data storages. Advantages of such arrays include: increased performance, fault-tolerance, and relatively low cost of disk space.

RAID 5 is commonly overrated to the level where it is considered the data storage, working and backup server all at once. Such opinion is a misconception, and may lead to further problems.

RAID 5 – is the array of 3 of more disks united in a single system, which writes the data to all disks in blocks. Along with the data, checksums calculated with a certain algorithm are written in a cyclic manner homogenously across all disks, providing RAID 5 fault-tolerance.

Everyone familiar with the basics of mathematical logics surely remember the simple operation called “addition modulo 2” (“exclusive or”, XOR). This very operation is the basis of checksum calculation algorithm, and it allows on-the-fly recalculation of the contents of any disk missing in array, using only the data and checksums stored on the rest of disks.

The total disk space dedicated to redundancy is equal to the capacity of one RAID5 disk. Correspondingly, the remaining disk space available for storage will be equal to (n-1)*V, where n is the disk count in array, and V is the capacity of the array’s smallest disk in GB.

Such system is implemented either with the use of RAID-controller (this solution is called hardware RAID 5), or by means of operating system’s tools, i.e. software RAID.

Level 5 RAID arrays can be different in several aspects:

• Disk count and capacity
• Disk order in array
• Array block size
• Data block writing algorithm (structure)
• Checksum “pattern”
• Presence/absence of shifts and discontinuities in data writes
• Presence/absence of RAID’s internal information areas

The procedure of raid 5 data recovery:

• The physical condition of disks is tested, and the set of acceptable operations is determined.
• Good disks are attached to the system bypassing any RAID-controllers, and excluding any possibilities of content modification.

Each disk’s content is analyzed in a HEX editor. Crucial points are located (MBR, boot-sectors, file system headers, partition boundaries), which allow identifying and confirming the information regarding the composition and level of RAID array in question.

As soon as we visualize the general picture of the puzzle, we proceed to assembling it. Of course, mosaic is best reconstructed from the structured pieces of a certain size. This can be a file allocation table or the data of particular types. The more experienced specialist is the more chances he has to successfully locate disk areas useful for analysis. At this stage it is possible to determine data block size, data structure, as well as checksum “pattern”. Also, shifts and discontinuities in the data stored on HDDs can be figured out.

Next, we find out whether the contents of operable disks are sufficient. If not, required missing parts are recovered. It is important to create the full image including every disk sector at this stage.

If we now have all the data on our hands, we can proceed to software assembling. That is, RAID-controller operation with the correct algorithm on the correct disk array is emulated by software tools. We arrange the disks in the correct order (while detaching unnecessary ones), and then adjust all parameters of the algorithm. It is not principal what soft exactly is used for this purpose.

Even if we get back (or recover) all the disks in array to normal operation, it is possible that we can still be unable to access particular blocks of data. As a rule, such discontinuities have cyclical nature and are caused by one or more incorrect procedures such as initialization, RAID rebuilds and so on. This is likely due to unqualified attempts of hardware RAID data recovery or “inconsistent” operation of the equipment in abnormal situations. Still, in many cases inaccessible blocks can be recovered by means of recalculation based on a certain set of other disks.

High reliability of storing data in RAID arrays is provided by means of data redundancy and implies fault tolerance towards failure of particular disks from a RAID array.

Increase in data access speed is possible only in some levels of RAID array and can be implemented by means of simultaneous execution of similar operations on different disks from array.

There are different levels of such arrays, for instance, RAID 0, RAID 5, RAID 10 and so on.

Unfortunately, regardless of all measures directed on improvement in RAID array data storage reliability, they fail to guarantee complete integrity of information. There are a number of reasons, which can lead to destruction of data. In general such reasons are related either to human element or to failures of hardware and software.

Let’s examine the ways of data recovery from RAID arrays of different levels in detail:

RAID 1 data recovery

1. First, analysis of destruction over all drives is carried out. Next, each drive is copied to an image file excluding defects. After reading all stable portions of data, problem portions are read, and finally the copy most relevant to the original is made.
2. Merging of two copies with regard to defects (subject to data relevancy to the original) is carried out.
3. If present, logical impairments of file system are fixed, and result is copied to another media.

RAID 0 data recovery

1. Analysis of destruction over all drives is carried out. Next, each drive is copied to an image file excluding defects. After reading all stable portions of data, problem portions are read.
2. Interlace block size is determined.
3. Offset for each drive and the interlace order are determined.
4. Result image is created, while fixing impairments in file system if needed. Then it is copied to another media.

RAID 5 data recovery

1. Analysis of destruction over all drives is carried out. Next, each drive is copied to an image file excluding defects. After reading all stable portions of data, problem portions are read.
2. Interlace block size is determined.
3. Offset for each drive, parity interlace block order, and the data interlace order are determined. If only one drive from array fails, it is not necessary to recover data from it, since the data from missing disk will be restored based on parity block and the data on the rest of drives.
4. Result image is created, while fixing impairments in file system if needed. Then it is copied to another media.

RAID 6 data recovery

1. Analysis of destruction over all drives is carried out. Next, each drive is copied to an image file excluding defects. After reading all stable portions of data, problem portions are read.
2. Interlace block size is determined.
3. Offset for each drive, parity interlace block order, and the data interlace order are determined. If only two drives from array fail, it is not necessary to recover data from those, since the data from missing disks will be restored based on parity block and the data on the rest of drives.
4. Result image is created, while fixing impairments in file system if needed. Then it is copied to another media.

RAID 10 data recovery is similar to RAID 0 data recovery with only difference that it may be necessary to determine relevancy of copies against originals.

RAID 50 data recovery is similar to RAID 5 data recovery with a difference that it will be necessary to recover two RAID 5 arrays and merge them into single RAID 0 image afterwards.

RAID 60 data recovery is similar to RAID 6 data recovery with a difference that it will be necessary to recover two RAID 6 arrays and merge them into single RAID 0 image afterwards. Matrix RAID data recovery is basically recovery process of one or more RAID 0, 1, 5 arrays depending on actual configuration.

Pioneer companies which offered offsite backup services made their appearance in late 1990s. Today the majority of online backup service providers use the SaaS (Software as a Service) model.

SaaS is a business model of selling and using the software such that the provider develops a web application and maintains it on his own, while customers purchase the right of accessing the software over the Internet.

The principal advantage of SaaS model for a customer is the absence of expenses connected with installation, updating and maintenance of hardware, and software running on it.

A payment is usually made in a form of yearly or monthly subscription fee depending on data storage capacity ordered.

Thus, offsite backup users are totally free from spending money on purchasing and upgrading their own storage systems. Using remote storage run by acknowledged, reputable providers may even significantly increase enterprise storage system security level in whole.

Online data backup services have a few common features. All of them are using Internet for transferring information to remote data storage, they suggest automatic backup while computer is idle, and allow scheduling backup creation during off hours.

System administrators can adapt data backup scheme for any particular machine. Recovery after disastrous situations as well as casual recovery from backups is also accomplished via Internet.

The distinction of offsite backup services lies in customer fee, speed and control options over remote backup modes. The majority of such services only transfer that portion of data which was changed since the latest backup has been made.

They also adjust data transfer rate so a backup service doesn’t take away the entire Internet channel bandwidth. The list of additional functions can be extended with data encryption, compression, and data deduplication options.

Here goes the list of the most popular offsite backup service providers.

MozyPro. The service is owned by Berkeley Data Systems start-up. The company was founded in 2005. Now MozyPro is used by over 3 million customers across the world including 70,000 corporations. Total data storage reaches 70 Petabytes spread among numerous datacenters over the world.

Arsenal Digital Solutions USA Inc. The company was founded in 1998 and had been specializing in data protection within distributed environments for a long time. Today it has storage of 20 Petabytes in 65 datacenters. The company’s major services are: ViaRemote – the service of data protection for PCs and servers (including VMWare virtual servers), ViaBack – for applications and servers protection, and ViaManage – for local data protection.

EVault Inc. EVault has purchased one of network backup pioneers – VytalNet Inc. provider – in 2001. It has since become one of leading online differential backup service providers. The company offers disk-to-disk backup service in network and local variants. They offer products both for desktops and servers. In 2007 EVault had been purchased by Seagate Company.

iDrive Pro. It is a service owned by Pro Softnet Corp., an ASP provider located in California (USA). It has more than 70 million customers and provides a total data storage of about 15 Petabytes.

SOS Online Backup. The company was established in 2001 in Australia, and then transferred their operations to territories of USA, Europe, and India. Business-product offered by the company is SOS for Servers. Company incorporates 11 datacenters across different continents.

AmeriVault Corp. The company was established in 1998 and had been specializing in online data backup and application management services. It provides services of continuous real-time data protection and duplication as well as automatic recovery in case of data loss.

Iron Mountain Inc. This provider offers services of PC and server backup, data recovery, as well as cataloging and searching of digital documents. It has 10 datacenters in possession.

Application management service toolkit in the cloud is meant to check services for availability, performance, and compliance with required parameters, as well as to monitor expenses related to deployment and provision of these services.

The key element of a cloud-based application management service is tools of automated detection capable of distinguishing and detecting changes within service components, including virtual and physical servers, network connections and applications.

In order to implement dynamic services, processes and tools of performance management and capacity management are required aside from virtualization and automation technologies on BSM (Business Service Management) level.

Performance parameters are stated in the Service Level Agreement (SLA), while software tools should provide a way to track the parameters in real time. That gives consumers control over actual performance.

Capacity management allows optimizing of cloud resources allocation and distribution according to the data on users’ specified demands and current changes of those. Capacity management also provides an option to plan allocation of needed resources for new cloud services (including cloud backup services).

One of crucial cloud features is self-service. A convenient tool providing such functionality is service directory, a well-known ITSM (IT Service Management) methodology element, which serves as a foundation for request management process in cloud environment. This directory is a “menu” of cloud services and it is usually complemented by self-service portal for users.

User’s request given to service directory can transform to one or more detailed queries, for example, to create a virtual machine or to install a corresponding application. After that, automation technologies should take care of altering cloud infrastructure “on the fly” to accommodate user’s request.

Configuration control process is based on underlying configuration database, also known as CMS (Configuration Management System), which displays infrastructure status and serves as the foundation for building service maps in BSM systems with the purpose of tracking technologic components’ influence on service implementation and, ultimately, the influence on business processes.

CMS should be capable of: integrating configuration elements of different origin – physical and virtual infrastructure components, private cloud elements, and external resources; using the tools of real-time automated component detection and dynamic propagation of changes.

For observing the SLA (Service Level Agreement) regarding cloud services, an availability management process is needed. For example, if a cloud service performance starts to decrease, the system will automatically allocate additional cloud resources to keep up with performance parameters stated in SLA. The process of managing SLA compliance helps to guarantee that cloud resources will meet security policies established on a customer enterprise.

One more requirement for a cloud-oriented application management service of new generation is worth mentioning.

The complexity of the environment makes demand for more convenient personalized tools for various roles in IT-service, which are going to maintain efficient interaction between specialists and serve as a unified platform for access and analysis of required information regarding cloud services management.

First of all, let’s figure out what is regarded as data reduction technology also known as Capacity Optimized Storage, COS or Capacity Optimized Transport, COT (applicable to data transfer via territorially distributed network channels).

In fact, data reduction consolidates fragments of the same content, providing for reduction of disk space required for storing the file.

Fragment size and optimization algorithms may vary, however every data reduction solution is based on the same principle, which implies decomposition of data to blocks of certain size and detecting repetitive blocks.

Now imagine how this technology works in regard to backup operations. When backup of files is created using data reduction technology, the resulting backup size is smaller than the original due to savings by merging repetitive fragments. Aside from the fact that disk space volume saved in this way can be rather significant, data reduction has one more important advantage. Identical fragments are copied only once. Thus, starting from the backup cycle number two, the system saves only a small part of original content to backup storage, because with the course of time the files are practically left unchanged for the most part.

Think about how many identical pieces of information are stored in your data center, and you will realize why backup applications based on data reduction technology can achieve amazing compress ratios – sometimes reaching 10:1 and beyond.

Compression provides for faster execution of backup cycles. Besides, usage of data reduction technology simplifies data exchange between distant offices, speeds up data transfer and cuts expenses spent on communication services. It is not surprising that vendors enhance their major products with replication applications.

Perhaps the most substantial effect of data reduction technology is the fact that the cost of backups stored on disks becomes quite comparable to cost of tape backups. Therefore, the price difference between these two media types given the same volume of stored information is almost neglected.

Competing manufacturers has been developing proprietary data reduction tools taking different approaches. Data Domain Company offers an integrated software plus hardware solution for backups, while Avamar is ready to deploy Axion application both in standalone product or bundled with required equipment. Maybe the most substantial difference between these two products lies in the fact that Data Domain’s product is integrated with basic backup applications, and Axion software just replaces them.

“DD460 is a fully functional disk system that serves as a network connected storage device for backup software”, ­­– Data Domain CEO Frank Slootman explained. – “Inside the black box there is our secret sauce, which provides for repetitive duplicates detection in different ways and their further unique allocation on the disk.”

“You can purchase our equipment or request the list of devices which were already tested by us and had shown a decent performance”, – Avamar CEO Ed Walsh stated. – “In general, Axion operates properly on any hardware running Red Hat Linux operating system.”

Taking into consideration that the volume of stored data gets doubled every year for most companies, data reduction system deployment may prove to be the most efficient way of maintaining systems’ readiness and retaining the controllable state of data protection tools.

Technologies of compression and deduplication play an especially important role in the reality of exponential growth of data volumes.

Data deduplication continues its extensive development in area of storage systems for backups, archives, virtual machine images etc.

Its usage not only allows saving storage capacity, but also increasing speed of saving and restoring the data. Various storage systems will soon have deduplication functions as a built-in feature.

First there ware magnet tapes, and it was the sole means of storing backups. Then capacious yet inexpensive hard disk drives appeared. Several years ago disk systems enhanced with data deduplication mechanisms were introduced. Such systems are capable of excluding repetitive duplication of identical fragments and now they hold an intermediate position between tape and hard disks.

Each of these technologies has their respective advantages and downsides. Fortunately, downsides of one technology can be compensated by advantages of others if we combine them all together in a D2D2T (disk-to-disk-to-tape) hierarchical system enhanced with aforementioned deduplication.

There are a lot of intuitive approaches to implementing data deduplication. It can be performed in arbitrary combinations on either file or block level real-time, or during further processing of copied data. Regardless of particular choice, the essence is basically the same: prior to saving a new data fragment, the system records its “fingerprint” using one of several algorithms. After that, if the system encounters the fragment with already known “fingerprint”, such fragment is not saved; just the path to its existing copy is saved instead.

Process reliability obviously depends on the degree of fingerprint uniqueness, which in its turn depends on the algorithm selected.

If we compare different approaches in terms of efficiency, then a usual compression with known archiving algorithms cuts the data volume to a half on average. File level deduplication used in CAS (Content Addressed Storage) allows cutting the volume to three or four times less, while switching to blocks or even smaller portions (called chunks) has potential of reducing the volume up to 20 times.

Technologies of block data deduplication are usually provided by companies specializing in virtual tape libraries, such as Avamar (purchased by EMC Corporation), Symantec Puredisk, Asigra, Data Domain, Diligent Technologies, Falconstor, Sepaton and Quantum. Network Appliance offers proprietary solutions as well.

Technologies of file data deduplication are offered by EMC in Centera product line, and by Hitachi Data Systems owing to purchase of Archivas and Caringo.

The maximum security level during data recovery can be provided by disaster recovery replication technologies, which make a copy of each new data record in the production system to the backup storage system located either locally or at a remote site.

There are two main types of IT disaster recovery replication: synchronous and asynchronous. Each corresponds to different levels of application importance.

Synchronous disaster recovery replication
Synchronous replication allows performing data backup to the secondary storage system each time when a write operation occurs in the primary source. This means that recovery point objective (RPO is the maximum amount of data which is tolerable for a company to lose) will be equal to zero, i.e. no data will be lost, and the normal operation of business applications on a backup site can be restored almost immediately.

However, such a high level of protection is traded for the inability of implementing the synchronous replication at remote sites.

This technology induces a heavy load on network connections. Besides, since each write operation waits for remote site’s duplication acknowledgement, it introduces delays in data transfer, which affect applications’ performance.

As a rule, these delays do not exceed an acceptable level only when the distance between the main and the remote site is not greater than a couple dozen miles or so.

Asynchronous disaster recovery replication
The problem of distance is no longer relevant in the case of asynchronous replication, which does not imply making backup of each single write operation. Instead, it puts these writes in a queue maintained on the main site and then uploads it to the secondary storage system once in a set time interval. Just like synchronous replication, asynchronous replication provides for a very quick restoration of system’s normal operation on the backup site.

However, storing a certain amount of data in the main site’s queue without actually making the backup increases the risk of data loss in case failure occurs.

Asynchronous replication is also prone to data integrity disruption, since the exact data writing sequence is not guaranteed to be identically duplicated to the secondary storage system.

Replication technologies allow implementing plans of BC (Business Continuity)/DR (disaster recovery) for the most critical business systems. As long as such systems are connected with networking in one way or another, they may prove to be more efficient when deployed along with global network data transfer optimization solutions, including integration of data deduplication and compression tools.

Today solutions for different types of data backup and replication are offered by virtually all leading storage system and network infrastructure vendors: IBM, HP, EMC, Hitachi Data Systems, Teradata, NetApp and others.

Among all developers of network optimization technologies we can highlight, for example, Riverbed Company, whose products provide for contraction of the size of data transferred across the network during backup or replication, thus reducing requirements for global network connections bandwidth, and also prioritize network traffic depending on the degree of data importance.

Imagine a secretary in search of a presentation, which was prepared the day before especially for meeting with clients. For some reason she cannot find it now. The needed data is not at hand, so the new presentation cannot be created, and what’s more important there is no time for that. If the enterprise lacks a clear tested disaster recovery plan, the following events are likely to proceed. The secretary calls IT administrator in despair only to be informed that he is on vacation. Priceless time is getting wasted, when she finally reaches someone from IT department, and again that coworker tells that he has only shallow knowledge about backup system and has never attempted emergency recovery. A half of the day will pass until he figures out how to recover needed data, then he may or may not recover the presentation in question, but it is too late anyway and the important meeting is already spoiled.

Now, if there was a correctly made disaster recovery plan, the secretary would certainly know who is responsible for handling such situations or the person who can substitute him in his absence, and she would call him directly. That employee, who knows the system well, would recover the missing presentation rather quickly, and the meeting with clients would proceed as planned.

Elaborately developed disaster recovery plan is urgently needed in case of complete loss of entire digital data. Efficient recovery and, as a result, quick restoration of current operation can be decisive for the enterprise’s further existence. That’s why disaster recovery plan is an obligatory element of any good IT security concept.

When making a plan, you should think about the following:

• For what purpose the enterprise is using digital data?
• When that data should be delivered?
• How would it affect the business process if the data becomes unavailable for one day or gets lost permanently?

IT disaster recovery team positions should be clearly designated. The most critical thing is to assign the person in charge of making decisions during disastrous situations and also his stand-in. If possible, one of managers should also be involved in planning so he will be able to assist in case of unexpected circumstances. It is important to inform all co-workers about disaster recovery plan and gather briefings devoted to it regularly. Also, it should be decided how exactly a notification about disastrous situation is made and who is the first person to be notified when it happens.

For events of total failure and serious disasters, a Disaster recovery plan should be stored in printed form at the office building and also at some safe remote location.

The issue about required quickness of reaction should be decided depending on the branch of enterprise activity and on the extent of how critical digital data is for enterprises’ operation. Should there be an immediate reaction in the event of data loss that occurred during off hours, or it can be postponed until the beginning of the next workday? Besides, the data should be categorized by its importance. The most critical data should be unquestionably recovered immediately while recovery of other data can be delayed.

Disaster recovery plan should contain the following information:

• telephone numbers and addresses of persons to be notified about the disaster;
• information on the person in charge of making decisions together with his stand-in;
• information regarding backup data center;
• information on IT service companies;
• address list of manufacturers and vendors of software and hardware.

Each stage of recovery plan instructions should be formulated and classified according to importance degree and extent of damage.

Along with instructions for regular employees, the document should include manuals regarding procedures carried out by recovery handling staff.

The document should contain step-by-step instructions regarding disaster recovery formulated in a way such that even a guest IT specialist is capable of understanding and carrying out all the steps.

Disaster recovery procedures should be written down with short, unambiguous and clear phrases. Graphics can be used in order to improve comprehension.

Practice alerts and trainings on disaster recovery plan usage should be conducted at least once a year.
After disaster recovery is fully completed, it should be checked whether all data is accessible again. Often some files are overlooked during recovery and users forget about them. In such case all further backups will not contain those lost files, thus exposing them to risk of being permanently deleted from previous backups due to Retention Policy.

Serious failures can happen anytime: even a smoldering fire at the data center is enough to paralyze its entire infrastructure. Aftermaths of floods are even more tragic, since whole server rooms permanently go out of service after being filled with water. The list of disasters can be continued infinitely.

In terms of handling such situations, medium business enterprises are prepared worse than others. And they need to pay special attention to two aspects – Disaster Recovery and Business Continuity.

The primary objective of IT disaster recovery is to restore and replace the damaged portions of IT infrastructure, as well as critical corporate data. As for Business Continuity, this concept involves many other aspects aside merely technical IT reconstruction: the idea is to prevent interrupts (or at least durable downtime) of business processes in case of a disaster, while allowing the enterprise to continue its normal operation.

The primary objective of Business Continuity is to minimize downtimes no matter what caused them – either a disaster or a scheduled maintenance of systems connected with taking them offline. By means of certain modifications and additions, corresponding methods and processes are appended to tasks of IT disaster recovery. For instance, in case of scheduled downtime of virtualized infrastructures, administrators can use systematic shut down of separate servers and storage systems in order to update operating systems and applications or upgrade hardware.

Among technologies used at times of unexpected failures are: Clustering, SAN (Storage Area Network) Replication, and also mirrored servers and storage systems in territorially separated data centers ­­­– all of these are needed for effective disaster recovery.

IT Disaster recovery planning

Medium business owners are challenged to answer two major questions: how to start solving Disaster Recovery problem and which technology is best suitable for this matter? Comparing to large companies, IT infrastructure of medium enterprises is seen as very compact. Nevertheless, first you should come up with a disaster recovery plan and then periodically test it.

A typical IT disaster recovery plan consists of hundreds of individual steps, starting from changing the cable connections and ending with detailed description of the order in which servers should be recovered.

However, having the plan itself is not enough – its efficiency should be proven in the course of regular testing. As a rule, individually developed means and recommendations come with an excessive number of complicated, manually performed steps that have to be taken in order to restore access to resources, recover operating systems and user data, as well as to test readiness of servers, network components and storage systems.

Instead of self-made solutions, it is reasonable to implement a virtual infrastructure along with specialized Back-up and Recovery software.

Shared storage system is a one of virtual infrastructure’s central components. The main objective is to store protected excessive local data here and create mirroring storage capacities available in some other place. A wide range of functions allow performing efficient remote data multiplication, IT disaster recovery and control over secondary storage systems.

Cloud computing has obviously aggregated so many various technologies, ideas, and products, that a certain portion of this ballast will have to be disposed of in the near future. As events evolve, it will become clear which technologies prove to be viable, and which fade out to the background of IT history. Anyway, we will try to make a brief overview of promising concepts and products, which have been attracting the largest amounts of attention lately.

Hybrid clouds
Hybrid cloud is an implementation of cloud technology such that one part of the system is located in public cloud, while the other part is in private cloud.

For instance, this kind of integration is possible when backup system is hosted in public cloud or vice versa – backup is done from public cloud to local servers. There are hybrid solutions that allow monitoring of both computing resources hosted within the company and those hosted on remote side (for example, CloudKick by Rackspace and System Center by Microsoft) at the same time. Another variant of hybrid cloud usage implies installing applications to the company’s internal servers, while renting additional power in the cloud owned by third party provider in case of unexpected load increase.

In the long run, exactly hybrid clouds are likely to become the dominating model of cloud computing implementation for organizations.

Computing power markets
One of cloud technology fundamental features lies in the fact that resources are available to a number of users at the same time, while being distributed among them as the load varies. The idea is that owners of private clouds, when being idle, can provide their computing powers for rent at an open market.

An opportunity to make some money is anyway more beneficial than letting servers stand idle. Besides, the open market would get new competitors, which would in its turn make for better pricing and quality of cloud applications.

For the first time such service was put into operation in February 2011 by one of cloud computing pioneers – Enomaly Company. With the help of their service, known as SpotCloud, the owners of surplus server powers (in order to participate an applicant should have at least 500 GB of free disk space) can sign up for free, install Enomally cloud controlling software to local computers and start earning rental profits.

Open source cloud platforms
Closed source code is one of major risks related to cloud technology: if a certain cloud provider becomes bankrupt or starts dictating unacceptable terms of usage, the customers literally will have nowhere to go.

Cloud platforms with open source allow solving or at least mitigating most problems connected with this issue.

Firstly, having source code on hands, any other provider is capable of bring the missing service online, which will also be completely compatible with the initial one. Thus, if one provider fails to continue operation, the customers’ businesses stability won’t be affected since there will be alternative providers ready to host “refugees”.

Secondly, the competitiveness of the market increases, which excludes the possibility of power abuse by monopolist.

Thirdly, the code being open source as a rule means that data formats are open as well: the customers can transfer their data from one system to another, or download it to a backup system anytime.

Fourthly, open source cloud basically allows setting up cloud solutions within the company’s own infrastructure. This option can be reserved for cases when there are concerns about excessive risk that is connected to using remote servers.

The first open source cloud platform to enjoy commercial success was IaaS-system called Eucalyptus, which evolved from a research project held by University of California at Santa Barbara. OpenStack project announced in summer 2010 loses to Eucalyptus in terms of maturity, but it greatly surpasses the older system in popularity. There are other open source IaaS-systems out there such as CloudStack by Cloud.com and OpenNebula. On top of that, VMware Company has introduced Cloud Foundry project in the branch of PaaS-systems.

Term cloud computing was introduced by Eric Schmidt, current executive chairman of Google. When he worked as a chief technology officer at Sun Microsystems, Schmidt already had come to a conclusion that network and computer are closely related. Later this concept spread around as official Sun Microsystems motto: “The Network Is The Computer”. That phrase is an utterly laconic version of the idea pronounced by Schmidt in 1993: “As soon as the network becomes as responsive as CPU, the computer as such will no longer exist, it will spread over the network.”

As of today, many interpretations of "cloud computing" term can be found. Each major analytic agency suggests their proprietary definitions, which often are ambiguous and complicated. Due to this, we think it will be the right thing to take the definition given by the National Institute of Standards and Technology as a reference instead.

A system is called a cloud system when it has the following features:

• Dynamic allocation of resources for the purpose of meeting users’ requirements related to computing resources and services
• Emulation of a “limitless” pool of virtual resources available on the first request
• Elasticity, which is namely an ability to dynamically perform horizontal scaling, both to the side of increasing computing power and to the side of decreasing it
• Wide range of ways to deliver computing resources and services to the end user
• Measurable quality of services provided. Maintaining consistent quality of service is achieved by automatic reallocation of internal resources.

A wide range of IT systems fall under this definition. Thus there is a problem of identifying an infrastructure or a technology as a cloud system.

The suggested solution lies in classification of basic services within the scope of this technology:

• SaaS — software as a service. It is a commonly used method of providing access to the software, which offers required functionality. At the same time the user has no means of controlling the provider’s internal systems. One exception can be made for software used for controlling access to the service
• PaaS — platform as a service. Cloud provider grants access to an application development platform within the cloud environment. The user has no means of controlling the platform or cloud infrastructure, but he can develop, test and execute own applications on the provided platform
• IaaS — IT infrastructure as a service. Cloud provider hands over a set of systems to the user. This set includes such systems as virtual server, data storage or network hardware. In this service model the user can install own software as well as control operating system, network connections, and data storage.

If we combine features of a cloud system and methods of service packaging, a question arises regarding the model of cloud ownership. Usually it is divided to the following classes:

• Public Cloud – this model implies that the infrastructure can be used by everyone. For example, such system us created by one of global providers, whose services are sold online and any user has an opportunity to buy the needed service
• Private Cloud – this model implies that the system is created and operated by a single organization and no one else
• Hybrid Cloud – mixed cloud ownership model. Such cloud implies that one cloud system (either private or public) is integrated with another cloud system (private or public).

Today tools and platforms for implementing cloud computing are offered by such companies as: Amazon (Virtual Private Cloud), BMC Software (Cloud Lifecycle Management), CA (3Tera AppLogic), Cisco/EMC/VMware (Vblock), Citrix Systems (Citrix Open Cloud), Cloud.com (CloudStack 2.0), Dell (Virtual Integrated System), Enomaly (Elastic Computing Platform), Eucalyptus Systems (Eucalyptus 2.0), HP (BladeSystem Matrix), IBM (CloudBurst), NewScale (NewScale 9), Novell (Cloud Manager), Oracle (Exalogic Elastic Cloud), Platform Computing (Platform ISF), Tibco Software (Tibco Silver), VMware (vCloud, vSphere and vCenter). Examples of complex (integrated) solutions are: vBlock by EMC, Cisco and VMware, FlexPod by Cisco, NetApp and VMware.

By definition private cloud is the infrastructure intended for usage by a single organization incorporating multiple customers (for example, departments of that organization), and possibly also by clients and contractors of that organization. Private cloud can be owned, maintained and operated by organization itself as well as by a third party (or a combination of those), while it can physically exist either within owner’s jurisdiction or outside of it.

Preconditions for setting up a private cloud

Some time ago, a platform could be in service for several years in a row, but under conditions of nowadays a system should by upgraded at least every year, which increases maintenance costs progressively.
Constant evolution of new technologies challenges every organization with certain difficulties. Small and medium business enterprises are not ready for continuous investments to IT infrastructure upgrading. Outsourcing is the only acceptable alternative under such conditions

Companies having IT departments with 10-20 year history do believe in the value of owning information systems and having full control over them. Companies, whose IT departments began forming a year or two ago, are not so conservative if the right to own turns out an expensive option. And the question is not about license costs. In order for an organization to host an information system, it should have a dedicated data center and the staff with corresponding qualification for its maintenance. Highly efficient and affordable system suitable for all relevant tasks of a business is not so easy to build. As soon as automation process will start within a company, IT department’s demands will be increasing in geometric progression, with all of them being important, absolutely necessary, justified, having a high ROI and also extremely high-priced.

What a company CEO should do in order to moderate the flow of IT demands?

The answer is not to build own data center.

This approach is quite usual for large businesses. Provided majority of offices are connected to data center not by local network connection but using some other means, why don’t move the data center to a professional operator? Professional operators provide environments where all technical requirements are met, emergency power supply organized, cold and hot corridors provided, a pair of reserve channels set up, security handled etc. Maintenance can also be delegated to third party organizations. As a result, practically all server hardware will operate outside of organization, while IT services, upgrades, adjustments and maintenance of both software and hardware components are carried out by outsourcing companies. Offices will house only workstations and a minimum staff of IT specialists responsible for tech support. But then again, tech support can well be outsourced, too.

If a concept of remote data center on provider’s territory is already acceptable, then it is time to take the next step – private cloud. Cloud services provide users with several collaboration models, when outsourcing depth is determined by the client himself according to assigned tasks.

- IaaS (Infrastructure as a Service) — both physical and virtual cloud infrastructures (networks, hardware, operating systems) are maintained by cloud service provider. At the same time, the customer is given an opportunity to install and run almost any kind of software including operating systems, platform software and applications.

- PaaS (Platform as a Service) — renting IT resources and operating system from provider. Customers get a flexible information encironment, which can be easily scaled according to tasks that need to be carried out.
- SaaS (Software as a Service) — is an opportunity to use any application without necessity to own a platform that runs the needed application. All updates, patches, hardware configuration changes are usually included to the subscription for chosen application.

Along with these types of of aaS, providers offer many other solutions: disk array as a service, business process as a service and so on. Absolutely every IT process can be outsourced. Today you don’t necessarily have to own an IT infrastructure in order to build an efficient IT environment.

The system of network backup consists of hardware and software components.
Hardware components are serving the purpose of writing and storing backups, providing a recovery source anytime it is needed.
Entire network backup logic is concentrated in software components and is implemented by them. They are responsible for management of devices, backup and data recovery processes, scheduling and also additional service functions.

Backup system hardware components

In order to choose hardware correctly, one needs to determine the media, which will optimally meet the requirements faced by the system, as well as the controller for this media.

Hard disk drives, magneto-optical disk drives and tape drives are commonly used as storage media.

A hard or magneto-optical disk would be suitable for one-time backup of a small-sized data array. A barrier for using hard drives is their high cost. If we’re talking about choosing a proper hardware for carrying out full backups, and in automatic mode, we should prefer tape drives: they provide the same recovery speed as magneto-optical drives, while showing a better write speed. The cost of data storage on magnet tape is the lowest comparing to that of other media. Therefore, the most efficient and reasonable storage for large volumes of data are magnet tape drives.

For working with magnet tapes, backup systems make use of so-called autochanger tape drives, or tape libraries.

Tape library consists of two functional devices:

- read/write device, or streamer, which is used for writing information to the tape and reading from it. At any given moment of time a streamer is capable of working with only one tape cassette;
- cassette autochanger device, or robot, consisting of a cassette stack and a mechanism, which inserts the requested cassette into the streamer and ejects it afterwards.

Usually tape library vendors only make robots and integrate them with streamers made by other manufacturers, selling resultant products under their trademark. All streamers currently manufactured implement read/write technologies providing high density of stored information, and as a result large capacity and high reliability of storage.

Streamers based on DLT technology, which is rapidly developing lately, are quite common now. The features of this technology are high write density, fast data transfer and reliability. All manufacturers of tape libraries generally use DLT streamers in their latest models. Other quite widespread standards of magnet data storage are DDS-2 and DDS-3.

Although the procedure of backup creation is quite simple, it can as well become a headache if it is not carried out regularly. Many businesses directly depend on manipulations with information, which is stored on their servers. Databases, source code repositories, web projects etc. All of those should be saved to backup media on daily basis.

In case the information is lost and cannot be recovered, a company may suffer a considerable damage.

Let’s examine two easiest schemes of creating backup copies.

1. Data copying from product server to backup server, where product server is a working server, which runs some services for users, and backup server is a server where product server data gets copied to. The sole purpose of such server is to store data copied from other servers. Typically it doesn’t provide any other services.

The main requirement for backup server is extensive disk space. Disk drives’ transfer rates are usually of no significance, since access occurs not too often: just at times when backup is written to disks and when backup is read upon relevant occasions.

The downside of this solution is the necessity to set up a dedicated server for backups. This obviously leads to additional expenses. However, smaller organizations usually tend to spend as less as possible on purchasing accessory equipment.

2. Data cross-copying is a method of backup creation such that two or more product servers copy their data to each other. This method is applicable if both product servers have sufficient disk space for storage of data from other servers. In such configuration product servers will act as backup servers to each other.

Data storage systems
Methods of organizing backups listed above are appropriate when it is a relatively small amount of data in question, within the range of a few hundred gigabytes. Otherwise, it is reasonable to consider a variant of DSS (data storage systems) application:

• Disk arrays - actually similar to backup servers, specially designed for data storage. Such arrays consist of multiple spacious HDDs.
• Tape drives (also known as streamers) - just like with tape libraries, data is written to special cartridges (magnet tapes encapsulated in plastic box).
• Tape libraries - intended for automated backup creation. Simultaneous operation of several taping devices increases the library’s performance and reduces the time needed to carry out read and write operations on backups.

It is also worth mentioning that online cloud backup services are attracting a steadily increasing demand these days. Such services are based on using remote backup servers operated by third party providers via the Internet. Particularly, more and more companies rely on a virtual private server (VPS), which combines functionality of both product server and backup server in one package. Such solution is fail-safe to a greater extent and is more economically reasonable, since it eliminates expenses connected with purchasing and upgrading the hardware, while providing an extra advantage, namely an opportunity to access the data from any point in the world.

Vendors of cloud backup services.

1. 3X Systems. 3X Backup is the main product of the company. It is an online backup service offered in a form of inexpensive portable configured device, which provides user with an additional local resource for backup and reliable data storage.

2. Amazon Web Services. AWS platform was the pioneer of cloud infrastructure and continues to hold dominative position at the market. Simple Storage Services (S3) are based on AWS and allow storing huge data arrays for a low price.

3. Asigra. A public cloud or a private one? Asigra offers both: Public Cloud Backup Service and Private Cloud Backup Solution, which provide for two variants of backup implementation without requiring agent application to run on your servers.

4. Axcient. The company offers data protection and provision of fail-safe business operation for IT managed service providers (MSP) in a form of hybrid backup device, which is installed at the customer’s site and is integrated with cloud emergency recovery service. Moreover, it can be extended if necessary without bothering spending money on infrastructure, licensing and software.

5. Carbonite. The company offers unlimited backup to the cloud for a fixed rate. Over 80 billion of files have been already backed up with the use of this service, and over 7.2 billion files have been recovered.

6. Cleversafe. Publicly accessible, private and hybrid clouds receive proper attention, either for data backup, archiving, content distribution or many other applications.

7. Doyenz. Offers a package of cloud backup services for IT-service providers, supporting reliable backup and transparent virtual failover. Company’s solution called Shadowcloud stores an active image of working server in the cloud, which can be called at any moment, making for fail-safe business operation.

8. eFolder. Offers high-reliability data protection services for both remote and local e-mail backup via extensive army of VARs and MSP. eFolder technology uses a method of reverse delta.

9. Intronis. Operates in cloud since 2003. Offers services of data backup and emergency recovery for MSP. The company uses AES 256-bit encryption and numerous Data Centers across both coasts of USA, providing for data protection and high level of readiness.

10. Mezeo. Offers a cloud storage platform, which is easily adopted, serves large number of customers, provides for high scalability and security. Its API interfaces — Cloud Storage API and Interoperability API — help service providers to monetize storage by using own products.

11. Nirvanix. Security, reliability and backup are the main points of the company’s offer. Its CloudComplete kit incorporates various options for cloud storage implementation, including CloudNAS Gateway, publicly accessible cloud called Storage Delivery Network as well as hybrid and private cloud solutions hNode.

12. Scality. Scality RING software allows transforming a typical x86-server along with an Ethernet network into a cloud storage system, while cutting the expenses.

13. Vembu. Enters the cloud storage market with a complete series of products and services related to online hybrid and cloud data storage: these are online backup service Vembu Pro, cloud data storage service Vembu Home, online backup software StorGrid SP Edition and hybrid backup software StorGrid Pro Edition.

14. Zetta. Offers corporative class NAS-system functionality in a form of cloud service on demand. Company states that its cloud solutions eliminate expenses spent for data protection and backup, provide for instant scalability of storage capacity, as well as keep an opportunity of extension in the future.

One of cloud technology elements that enjoy a wide implementation in the real life is remote backup, also known as cloud backup or online backup. As of today, tens of cloud backup services offer cloud storage for backups under various conditions and fees. Cloud backup market is extending with rapidity of geometric progression.

Cloud offers a set of advantages compared to usual (local) backup creation.

Advantages of cloud backup
Firstly, stored backup can be accessed from any location in the world. To a certain extent cloud backup is another way of data transfer both between computers of the same user, and between different workstations.

Secondly, user no longer has to care about integrity of a compact disk containing important backups or viability of a volume on hard disk drive dedicated to backups. Since data is uploaded to the cloud, its physical preservation becomes responsibility of the data center’s owner, whose service is used by cloud storage provider. The portion of liability is usually denoted in user agreement.

Thirdly, cloud backup is traditionally offered as not just a free space on remote disk, but as a tool of backup automation. Customer receives client software for one of popular platforms, as well as access to administrative interface to his personal section at the server, which allows adjusting settings of backup to meet even the most exotic needs.

However, regardless of obvious advantages, many potential customers are rejecting the service due to trivial distrust to cloud technologies, particularly connected to the danger of third parties getting access to the information. As a result, a portion of customers refuse to upload their data to the cloud, and another portion can accept it only for backup of non-critical data (and immediately a question arises, where to put critical data then?).

Up to recently, low bandwidth channels were also listed as a disadvantage of cloud backup. But the largest amount of information is uploaded to the server only the first time it is accessed, and only changes are uploaded afterwards. Results of a day-long work of most users can be easily transferred to the service even if the channel is not really fast. This disadvantage gradually vanishes as the Internet is developing though.

We can tirelessly speak about the necessity of information backup. But, unfortunately, the strongest motivation for taking backup seriously arises mostly after incidents of serious loss of critical data.

Backup creation methods

The easiest and clearest method of backup copying is so called Full Backup.

The essence of this method is taking all our data and making its physical copy. As soon as Full Backup is completed we will have two copies of our entire data: one is the original, and the other is the backup. Both copies will be identical. The downside is that full copy will occupy the same amount of disk space as the original; besides backup creation process can take a considerable period of time.

Due to this reason, two other methods of backup are used more often: Differential backup and Incremental backup. These two types introduce a combination of periodic, but not frequent, backup creation, and creation of differential or incremental copies in between of full copies once an hour, for example. Differential and incremental backups are created quickly and consume significantly less disk space compared to full copies, since they do not include all files, but only those created or changed after the latest full backup.

Differential backup
Let us examine the method of backup called Differential backup in detail.

When it is used, only those changes are backed up which occurred after the latest Full Backup. Since differential backups can be created either on image level or file level, this set of changes will be presented in a form of altered disk blocks (for image-level backup) or changed files collection (for file-level backup).

Major benefit of differential backup lies in significant decrease of time needed to create the backup compared to Full Backup. On the other hand, recovery after a failure takes more time. Recovery process will require two data restoration operations to be performed. During the first operation, the data from Full backup will be recovered, while the second will recover data from differential backup.

When used for inexpensive data storage subsystems, file-level Differential backup is selected in cases when applications create numerous small files and alter only small subset of those files after Full backup is created. However, this backup method is not used in case the hard disk drive is used by Database management applications, which constantly make changes in huge database files. Thus, in case of file-level backup, a backup copy of the huge will be created over and over. Microsoft Exchange is an example of a program which tends to constantly alter the contents of huge database files.

Examples of Differential backup implementation
Let’s consider a few real life examples to demonstrate how Differential backup technology is implemented in practice:

Files and directories. Suppose we need to back up a directory, which contains 50 files having the size of 1 MB on average. Accumulated changes in files comprise 2 MB per day. Creation of backups of all changed files will require 50 MB of disk space per day. Using differential backup will allow decreasing this amount to 2 MB, then 4 MB, then 6 MB, that is no more than 2 megabytes per day, exactly the amount taken by changes themselves.

Disk images. Differential backup is very efficient to create a disk image and allows saving disk space. However, to compute the difference Full backup still needs to be created, and thus it requires corresponding volume of free disk space, which is not an option for many users.

Databases. Backup of databases is the sphere, where Differential backup is extremely important. A typical backup plan used by modern companies requires a weekly backup of Exchange Server. Instead of copying tens of gigabytes of data, Differential backup allows decreasing backup sizes to hundreds of megabytes and only make backups of new correspondence, which appeared after the latest Full Backup took place.

Storage Area Network (SAN) is a special dedicated network, which connects storage media with application servers. It is usually set up using Fibre Channel protocol or iSCSI, which has been picking up a lot lately.

Reasoning behind creation of Storage Area Network (SAN) was in part similar to that formerly seen behind usual LAN network with shared devices in it. If we buy a laser printer for every client computer, all printers will probably be idle 99% of the time. Instead, we can be better off installing one shared printer on the network. Even though a shared printer can be more expensive than a personal one, in total it is still cheaper because you only need one, besides such printer has better functionality and reliability.

Along with SAN there are also two mainstream methods of data storage systems organization:

1. Direct Attached Storage – a solution, which designates that a storage media is directly connected to the server or workstation. As a rule, Fibre Chanel, SCSI, eSATA or SAS interface is used.

2. Network Attached Storage – a standalone disk system, acting as a dedicated server running a specialized OS, which exclusively serves functions related to file system maintenance and input-output implementation. NAS incorporates a speed-optimized file system. NAS devices are designed in a way such that all their computation power is focused on operations of file storage and file system maintenance.

In opposite to NAS, SAN has no idea about files: file operations are executed on servers connected to SAN. Storage Area Network performs operations on blocks as if it were a sort of large hard disk drive. An ideal result of SAN implementation is the availability of any part of storage capacity within SAN to any server running any operation system. End elements of SAN are application servers and storage systems (disk arrays, tape libraries etc.) Similar to regular network, adapters, switches, bridges and hubs connect all elements together. iSCSI is a more “friendly” protocol, since it is based on standard Ethernet infrastructure: network adapters, switches, cables. Moreover, exactly iSCSI-based storage systems are most popular for virtualized servers owing to simplicity of protocol setup.

Storage Area Network advantages:

• High reliability of access to the data located on external storage systems.
• Independence of SAN topology from data storage systems and servers.
• Centralized data storage (reliability, security).
• Convenient centralized control of commutation and data.
• Decrease in LAN load due to moving intensive input-output traffic to a separate network.
• High performance and low latency.
• Scalability and flexibility of Storage Area Network logical structure
• Opportunity of organizing remote backup data storage systems and remote system of backup and restore.
• Opportunity of constructing fail-safe cluster solutions based on existing SAN without additional expenses.

Virtual networks
As a concept, virtual network should be considered as a switch, but in opposite to a physical device it is a virtual switch.

All virtual machines within one network are connected to one switch. Each virtual switch is logically isolated from the rest of virtual switches. If there is a need for nodes connected to one virtual switch to connect to nodes connected to another virtual switch, a virtual router, VPN server, virtual firewall or similar device should be set up, just like for physical network.

It is very handy to create different virtual switches if there is a need for network isolation similar to isolation that exists in a physical network, where different segments are separated with a firewall or another access control device.

Virtual network types
There are three types of virtual networks, which can be connected with Hyper-V:

1. Private Virtual Network
2. Internal Virtual Network
3. External Virtual Network

Private Virtual Network

Private Virtual Network is a part of network virtualization. It is a virtual switch, which connects virtual machines only. Guest OS’s connected to one segment of virtual network can interact with each other, but cannot interact with Host operating system, and similarly host operating system cannot interact with virtual machines on the private virtual network.

Private virtual networks are extremely handy if you are aimed at complete isolation from the rest of virtual networks, but sometimes this approach can introduce problems if you need to copy files to virtual machines since virtual machines cannot connect to underlying (host) OS. In this case another virtual machine should be set up, which will take the role of a router or a firewall. Then, one virtual network adapter should be connected to the private virtual network, while the other should be connected to the physical network (via an external virtual network).

Internal Network virtualization

Internal Virtual Network is similar to private virtual network in a part that it is not bound to any physical NIC. Internal virtual network is also an isolated virtual switch like in private virtual network, but in case of internal virtual network Host OS has access to guest virtual machines via virtual switch of internal virtual network.
However, these virtual switches lack DHCP function (which VMware Workstation offers in some cases). Therefore, if there is a need of interacting with virtual machines connected to internal virtual network’s switch, you will have to allocate IP address for a virtual NIC of Host OS connected to internal virtual network such that address is also valid for the virtual network to be interacted with.

External Network virtualization

External Virtual Network differs from other types of virtual networks, since this type is bound to physical network adapters. You can have one external virtual network per each physical NIC installed to your Hyper-V server. External virtual network’s switch will be treated in a way as if it was a physical network adapter of the server. That’s why if you look through configuration of the previous NIC, you will probably think that it has no IP addresses configured. Instead, virtual NIC is added in Network Connections window and this virtual network adapter is connected to external virtual network’s switch. Besides, IP addresses allocated to this virtual NIC provide for its interaction with physical network.

As a result of efforts of various software companies and developers offering this service, several names were established for the same technology:

1. VPS - Virtual Private Server
2. VDS - Virtual Dedicated Server
3. DDS - Dynamic Dedicated Server

What is VPS after all?

Long story short, VPS stands for splitting one physical server to multiple separate independent virtual servers.
Both Unix and Windows systems can be split. As a result of separation, a few standalone dedicated servers are formed, while still being physically located on one server.
Each VPS has its own operating system distribution kit, in which it is granted practically unlimited rights of super-user. This means having root (UID: 0) level rights under Unix or Administrator account under Windows, which allows compiling or installing specific software with specific configuration.

VPS advantages
Each VPS can have one or more IP addresses, through which inbound and unbound traffic goes. Even if one VPS happens to be under DDoS attack, other ones won’t be affected (it’s worth to mention that other VPS can still be affected, but due to secondary reasons, for example in case the server’s shared channel is not broad enough and attacking packages fill it up completely).
VPS guarantees that minimum resource consumption will be observed. Client will never face accusations considering excessive usage of CPU or memory resources, which often happens on regular virtual hosting providers.

Quick reboot of virtual server, opportunity to back up VPS by taking a snapshot of the entire system.

Increased stability of operation for lower expense.

Virtual private server is suitable for projects having a high visitor count, but not requiring large portions of memory and CPU resources constantly. VPS enjoys the greatest popularity among companies’ official websites and online shops.
There are several mainstream VPS-hosting technologies: Xen, VMware, Virtuozzo.

Xen and VMware technologies allow creating a virtual environment, which completely imitates physical server operation. This excludes a possibility of one server interfering with another, regardless of consumed resources portions. Besides, Xen and VMware technologies provide support for almost any operating system.

Virtuozzo technology allows distributing physical server resources within one or another operating system. The main difference of this technology from latter two is that it can provide clients with a greater amount of resources than actual server has. This allowance relies on the fact that the entire server resources are not likely to be demanded simultaneously. Therefore, this service is practically identical to a regular virtual hosting. Virtuozzo technology does not provide isolation of virtual servers one from another, which means that resource-heavy virtual machines may interrupt operation of other clients’ servers.

Usually desktop virtualization is understood as a technology that provides user with full access to their working environment running in the server’s virtual machine. Virtual desktop is in fact an image of a computer, which contains operating system, applications and settings. Client machine used for access to virtual desktop image only needs to provide remote connection to the server, which means that user is no longer tied to a certain PC. With desktop virtualization user can work with their files and applications located on the server side from any terminal, including territorially remote ones.

Such way of connection provides for an opportunity to access user’s working environment from any geographical point of the world, while the environment keeps the state, in which user left it after the latest working session. Working session virtualization (user commands, desktop display) is carried out by software of a thin client.

Thin clients
Thin clients connect to server, where all needed programs and applications reside. The same server provides access to data storage and carries out all computations.

Owing to thin clients implementation, load distribution falls completely on terminal server, and every user is working on a separate input-output device (terminal station), which doesn’t perform any computations itself and is merely used for determining the tasks and receiving server’s response.

Terminal stations can work well without major pieces of hardware or software, which are considered crucial for typical computers.

This technology introduces several obvious advantages.

Firstly, it makes user free from being tied to a certain PC, providing usual informational and computational environment anywhere and anytime.

Secondly, it allows considerably lowering the expenses related to the maintenance and support of numerous workplaces, since user stations become virtualized. Only one central server needs to be maintained including such works as periodical installation of new versions of system, client and applied software, as well as regular scheduled maintenance.

Thirdly, one more substantial advantage of desktop virtualization is the ease of virtual machine cloning. For example, a company establishes a new office, where 20 new workplaces should be organized for workers. Without using virtual machines, technical staff should have assembled and test computers, and then installed operating system and applications for twenty times.

Now, with desktop virtualization, it is only necessary to create one workstation and make 19 clones of it, after which only minimum tweaking of some virtual desktops is needed in order to meet requirements of certain users.

The same applies to data backup. If saving a backup of a regular physical computer is either only partially possible or requires bringing the system offline, then creating full backups (snapshots) of virtual desktops is possible at any time without interrupting the work. Restoring virtual desktop from backup is carried out multiple times faster than restoring backup of a real computer.

There are several forms of organizing storage.

DAS (Direct Attached Storage, also known as SAS, Server Attached Storage) – a classic way of organizing storage, such that a drive (disk array under a RAID-controller) is either located inside the server box or is assembled as a standalone device connected to one or more servers.

NAS (Network Attached Storage) – these are in fact autonomous file servers having a RAID-array on board (sometimes additional storage media can be connected to extend NAS volume). Usually integrated Web-server or SSH-console is used to control NAS.

SAN (Storage Area Network) – the most versatile solution for storage virtualization at an enterprise of any scale. SAN is a standalone network separated from the main enterprise network, which is commonly based on a Fibre Channel or iSCSI. Fibre Channel (FC) connects storage media and servers to each other. Lately Internet SCSI (iSCSI) is often used as internal SAN interface for the reason it is less expensive and allows virtually unlimited distances between network nodes. Data transfer between SAN and servers connected to it goes on block level, similar to DAS.

Data center virtualization technologies

There are two main approaches to virtualization technologies on SAN level: in-band storage virtualization and out-of-band storage virtualization. In the case of in-band virtualization, a special device is set up between servers and storage modules, which incorporates rules of converting physical storage resources to logical ones. As for out-of-band virtualization device, it is connected directly to server or switch/router and manages the information about data transfer between application and server on the way to storage media connected to SAN.

Server level virtualization
Historically, server-side storage virtualization was the first solution. Logical volume managers first made their appearance on mainframes, later on Unix-servers, and in later years they were introduced on Windows platform. They provide for virtualization by means of displaying physical devices as logical ones having so called logical numbers (LUN) and divided into logical disk groups or logical volumes. This allows applications to mount logical volumes without establishing a dependency on a certain physical device.

Storage subsystem level virtualization
Storage Virtualization Managers, which create Virtual Volumes, are used in mainframes since 1990s and to date. Another approach is called “SAN-in-a-box”; it is based on an integrated solution such that one stand is shared by drives, control systems and switches. This solution satisfies the goals of virtualization, however only within the “box”.

Storage network level virtualization
Any storage network consists of two component groups: functional (servers, drives) and infrastructural (adapters, hubs, switches). In order to implement storage network level virtualization, the third component group should be added, which may be called a controlling group. Devices that make the controlling group are called SAN-add-ons or SAN-servers. These are computing devices connected to SAN or installed on data transfer routes, which are responsible for topology and implement abstraction of the data from its physical location.

Dozens, if not hundreds of various applications may be needed by an arbitrary user in the process of working on a modern computer. Besides the effort of installing them on each user’s computer, further maintenance is required such as updating it on time, running different versions concurrently, providing for compatibility with other installed applications, and finally removing those programs once they are no longer needed (sometimes complete removal cannot be fulfilled and a portion of folders/files as well as registry keys created by the application remain intact). All these problems can be solved with application virtualization.
Let’s study an example with Microsoft Application Virtualization (App-V)

The idea behind this technology is as follows: instead of installing the application to the target computer, it is installed to a specially prepared “clean” machine (a virtual machine with client OS, for example). During the stage of application’s initial setup a special utility (that goes by the name Sequencer in App-V) tracks down every dependency of this application: files, folders, registry keys and branches, system libraries; and creates a package containing all of those. Packed application is then transferred to the target computer, where it runs inside an isolated container (also often called sandbox). User doesn’t have to install the application and adjust its settings ­– it just can be started without delay. At the same time user still has an opportunity to change various options of virtual applications, and set associations for file extensions.
There is an option in the centralized infrastructure of App-V to set rights for applications, allowing them to be run only by certain users or groups. Having started an application once, a user can log off the corporate network and leave for a business trip, and that application will run from user laptop’s local cache without the need of connecting to App-V server (or it won’t run if security standards of the enterprise do not allow that).
App-V virtual application packages can be distributed using the tools of application deployment to workstations (SCCM 2007, for example). App-V perfectly handles solutions for terminal access and presentation virtualization (based on MS RDS or Citrix XenApp) as well as VDI (VMware View, MS RDS, Citrix XenDesktop). In the latter case a considerable amount of disk space can be saved on virtual workstations, if all clients will have access to a read only cache for virtual applications located in a shared folder on the network. Taking into account that installed applications may take up to 20 GB of disk space, savings prove to be rather significant.
Finally, license purity can also be controlled with the help of App-V by setting the limit number of concurrently running instances of an application.

App-V Components:
Application Virtualization Client – a component installed to a client workstation or terminal server that allows retrieving, updating and running virtualized applications.
Microsoft Application Virtualization Sequencer – a component installed to a reference computer, where initial installation, setup and packaging of applications are carried out.
Application Virtualization Management Server – a centralized server, which allows publishing virtualized applications, setting access rights, managing licenses and much more.
Application Virtualization Management Console – a snap-in, which allows connecting to and managing one or more App-V Management Servers.
IIS Server – allows connecting to management servers of App-V Management Server with the App-V Management Console.
SQL server – used for storing management servers’ databases.
Streaming server – allows deploying application packages to client computers.

However, data storage system transformation and data center virtualization is not a simple task even within one data center, especially if we take to consideration the fact that IT-administrators try their best to perform all works without interrupting the enterprise activity. Nevertheless, problems can be easily solved if you stick to 4 simple rules described below.

Simplification of virtual data storage system
First rule. Prior to beginning data center transformation it is useful to evaluate the possibility of data center virtualization. If virtualization requires creation of additional control layer for association of physical and virtual resources, single points of failure, or binding to a certain provider, or structured zones, then it will only make already complex data storage network even more complicated. The simplest and most efficient solutions are those utilizing existing system controllers for standard connection to external storage media. This allows controlling external data the same way internal data is controlled, while supporting functional capabilities and performance of given controller.

Internal and external extension
Second rule. In order for extension to make for maintenance expenses reduction, unprecedented scalability opportunities are needed. You will be better off using such solutions that are capable of extending according to system consolidation requirements demanded by more and more powerful server clusters, as well as aiming at dynamic allocation of virtual resources for data storage on a growing number of servers and applications.

Built-in security system
Third rule. With data center virtualization, IT-administrators should provide for mechanism of protecting the data from unauthorized access. Ideal data storage system security can be provided by delimiting user zones and separating control data from user data, as well as by assigning different addresses to virtual ports, which rely on the same physical ports. However, for efficient operation of these security functions they should be initially built-in to products, and not be postponed for later implementation.

Transparency
Fourth rule. Applications should “see” both virtual system and the underlying physical infrastructure. Besides, data storage system should incorporate an opportunity of tracking service applications operation. The latter cannot be implemented without an integrated software tools that provide collecting data from the entire infrastructure, identify it with a server or an application, and finally output it to intuitive control panel.
These four simple rules allow creating a data center virtualization, which will provide global consolidation of resources, technologies and applications. Such virtual system simplifies IT-infrastructure and provides for protection of all data, which increases its significance for enterprise operation.

One more server virtualization benefit is software expenses reduction. Some software vendors have implemented separate licensing schemes especially for virtual environments. For example, buying just one Microsoft Windows Server Enterprise license you receive the right to use it on 1 physical server and 4 virtual ones (within one server). As for Windows Server Datacenter, it is licensed for a certain number of CPUs and can be used in unlimited number of virtual servers.

Maintenance expenses reduction is the next server virtualization benefit. Maintenance of lesser amount of equipment is cheaper and easier.

Electricity expenses reduction. The number of physical machines that work as servers is reduced, which lowers the energy supply. On average, servers are idle 70% of time, while still consuming energy and releasing large amounts of heat. Server virtualization allows spending CPU cycles more efficiently and rise utilization to as high as 90%.

Infrastructure flexibility enhancement. Virtualization allows software to abstract from hardware. Therefore, it introduces the opportunity for virtual machines to migrate between different physical servers. Earlier, if a server started malfunctioning, it was necessary to reinstall OS and restore all the data from backup on that server, which took hours. Now it is enough to perform virtual machine migration from one server to another, no reinstallation involved. Virtual server also demonstrates lower time of deploying solutions of your tasks owing to server system’s shorter installation, setup and delivery procedure.

Increase in level of fault-tolerance. Virtualization offers means to clusterize the entire server regardless of running software. Even unclusterizable services can be clusterized, too. In such case clasterization in its pure sense doesn’t occur; in fact it is just a virtual machine restarting. But, should a physical server fail, virtual machine will be automatically started on another one without system administrator’s attendance.

Last but mot least servrer virtualization benefit is that testing made simpler. Virtual servers offer a convenient environment for testing software, modeling loads, processes, and tasks. Even if something goes wrong or deviates the expected scenario, only the virtual environment itself will be compromised, while the physical server uninterruptedly runs other virtual machines.
Besides, when testing is performed on virtual machines, they can be migrated from work servers which are exposed to bigger loads to other physical servers.

Virtualization is the technology that allows making computing resources autonomous and mutually independent. It should be applied to each single level of computing structure, starting from data center and ending at standalone PCs.

The purpose of virtualization is replacing rigid binding between different levels, such as between operating system (OS) and devices, between application and OS, between user interface and local computer. Instead, virtualization makes dependencies between mentioned elements relatively loose. This introduces new opportunities, and sometimes problems. For instance, by using virtualization of hardware (including server virtualization and desktop virtualization) we can make OS independent of hardware, which runs it. Likewise, virtualization of applications can perform similar separation of OS and applications being executed within it. In a similar fashion, virtualization of presentations is able to untie user interface from physical computer, which runs the application providing this interface. While you are working with an application on one machine, it can be controlled from another.

Separation of these levels makes for a much greater flexibility in every aspect of IT-infrastructure. More various ways of both hardware and software usage are introduced, as well as opportunity to replace components more easily. Owing to isolation of critical portions, systems become more protected. Many tasks of IT staff become simpler. Unsolved problems that irritated IT staff every day are gone.

Virtualization is a powerful tool of infrastructure transformation that affects all levels from data center to desktop workstations. It offers benefits such as faster application deployment; uninterrupted availability of systems, applications and data; minimizing difficulties connected with servers and desktops recovery or taking them offline for testing and development; lowering costs and improving flexibility of the entire environment. You can also see the list of server virtualization benefits.

Integral concept of IT virtualization
Evidence suggests that large-scale virtualization can offer considerable advantages. However, it also became clear that virtualization has its specific difficulties, for example the necessity of maintaining “the islands” of overgrown virtualized and physical assets. That’s why it is recommended to use time-proven tools in order to make the task of controlling the entire infrastructure and all associated processes simpler and faster. Transition to IT virtualization should be implemented in a way that won’t lead to consequences negating its advantages.

This integral approach at virtualization is not just a theory. This is reality for Microsoft Corporation and most of their customers. In order to bring all-round virtualization to life, the corporation has united resources of specialists, partners, products and solutions.

Integral approach at virtualization will aid organizations in advancement towards dynamical condition, making for enhanced flexibility of IT systems and simplicity of changes.

Creating a virtual environment

Many system administrators choose to set up multiple servers. This is for a reason: it is recommended to devote different servers to different tasks. For example, Microsoft urgently advises to never run both Active Directory domain controller and Internet gateway on the same physical server, since it introduces a security threat. In case of malware or hacker attack on the network, Internet gateway is first to take the brunt. And if the domain controller shares the server under attack, AD bases will probably get corrupted or even worse – leaked to hackers. While outcome of corruption is just spending some time on recovery, in case leak happens more elaborate attacks are very likely to come using logins and passwords of actual network users. The least bad thing that can happen is that all company users’ e-mail addresses make their way to spammer bases.

With respect to the above, domain controller is installed to one physical server and Internet gateway is installed to another. This is correct. Now if hackers deploy a successful attack on Internet gateway, chances that they will reach anything past the gateway will be much lower.

Every separate server has its cost. If we’re speaking about brand servers here, the costs fly up. Each server consumes electric power, occupies place on a table or a stand. Besides, server applications rarely consume large portion of system resources: CPU load doesn’t exceed 10% for domain controllers and Internet gateways that we mentioned. Thus using a separate server for each of these tasks is seen as irrational. On the other hand, using a single server is unacceptable in terms of security. Well, where is the happy medium then? The answer is given with server virtualization .

What is server virtualization?

Server Virtualization is a software technology of computer’s hardware emulation. One physical computer can run multiple virtual machines at once. Such virtual machines act just like separate physical ones; you can install operating systems and applications to them. Each virtual machine uses a portion of physical host’s resources for its operation. And the portion of resources reserved for use can be adjusted either statically or dynamically. This technology allows a more rational usage of hardware resources.
Example: we have two applications that require 128 Mb RAM each and cannot be installed to the same physical server.

Our options are:

1. Buy two servers having 256M RAM each;
2. Buy one server having 512M RAM and run both applications within separate virtual machines;

Obviously, the latter option makes for more effective hardware resources usage (CPU in particular), while the cost of server virtualization is much lower, since one server having a larger RAM capacity is always cheaper than two servers. See all server virtualization benefits.

Virtualization is a proved software technology that transforms IT-environment at a growing rate and makes fundamental changes in computing processes.

Modern computer equipment has rather broad capabilities, but initially it was designed for one operating system and supposed to run only one application. But virtualization overcame these limitations and enabled launching several OS and multiple applications on a single PC at the same time, thus increasing hardware performance and flexibility.

IT virtualization gives advantages to any PC user. Millions of people all over the world use virtual technologies to save time, money and electricity, and in addition to that achieve better results without hardware extension.

Nowadays any large company that claims to hold leadership in its branch can hardly do without virtualization.

The following types of virtualization can be marked out:

• Server virtualization - launching of several virtual servers on one physical server. Virtual machines or servers are applications launched on hosting OS that emulate physical devices or servers. Each virtual machine can run on a different operating system with applications and services installed on it. Typical examples are products like VmWare (ESX, Server, Workstation) and Microsoft (Hyper-V, Virtual Serer, Virtual PC).

• Application virtualization means emulation of operating system's resources (register, files etc.). This technology allows using several incompatible applications simultaneously on a single computer, or to be more precise, on a single operating system. Application virtualization is implemented on the basis of Microsoft Application Virtualization product (AppV). AppV allows users to launch the same previously configured application or group of applications from a server. Moreover, all running applications will work independently, without making any changes to the operating system. All this happens transparently for the user, as if he works with a usual locally installed application.

• Presentation virtualization means emulation of user interface. I.e. user sees an application and works with it on his device, though in fact the application runs on a remote server, and only the image of the remote application is transferred to the user. Depending on mode of operation, the user can see the remote desktop and application launched on it, or just the application window. This is implemented on the basis of Microsoft Terminal Services and Citrix solutions.

• Operating system-level virtualization means isolation of services within one kernel specimen. It is implemented on the basis of Parallels (Swsoft)  or Virtuozzo and is most often used by hosting companies.

Choosing a right storage virtualization solution for your business can be very confusing. Different types of Storage Area Network (SAN) have their own pros and cons. Which technology to choose depends on particular business needs and how much you can afford.

But first it is important to know what makes the difference and what it means to you. Here we'll give you some ideas about two SANs that represent two opposite approaches to storage virtualization. Each with its partisans and opponents, iSCSI and Fibre Channel (FC) cause hot discussions in professional circles. Let's investigate and consider advantages and disadvantages of each.

Advantages and disadvantages of iSCSI and FC

Any SAN uses Hot Bus Adapter (HBA) to connect server to shared block storage array. As the result of such connection, storage network is established that enables a server to exchange data with storage. Various methods can be used to secure storage avaliability and improve its productivity.

Fibre Channel can transfer data using both twisted pair copper wire and optical fiber cables at a speed of 4 Gbit/s. FC is recognized as a «traditional» SAN, but many IT managers now start giving preference to iSCSI for it is less expensive and complicated. Let's see: you need HBA cards connected to a corresponding port for each server to implement Fibre Channel. This can cost you a considerable amount of money. If you are a small or medium-sized business, such costs can be beyond your strength.

The situation changed in 2003 with ratification of iSCSI protocol based SANs which allowed to send SCSI commands to storage on remote servers. Unlike Fibre Channel, iSCSI doesn't require special cabling. Ethernet LAN technology used by iSCSI is comparatively cheap and easy to learn. At the basic level, iSCSI can produce 1 Gbit/s Ethernet connection that actually can be upgraded to 10 Gbit provided network adapters and switches are adjusted to relevant speed.

However, some believe that iSCSI has low performance when compared to Fibre Channel and thus keep away from deploying this SAN. Experts maintain that iSCSI SANs are likely to face oversubscription and thus become unable to sustain the load. That's why it is recommended to use high-end Ethernet switches for iSCSI networks.

By the way, some FC devices can maintain 10 Gbit speed already and if it goes on this technology will probably attract more users.

Does performance really matter?

According to stats, performance is not a determinal factor when choosing between Fibre Channel and iSCSI. There's a considerable difference in approaches to administration and management of these SANs. As mentionedabove, FC requires peculiar skills and iSCSI can perform poorly on existing TCP/IP protocols. As of now, both technologies remain in parity. And if you have a Fibre Channel SAN, there's no reason why others should not deploy iSCSI and visa versa.

URL - Uniform Resource Locator

URL Encoding

Try It Yourself

URL Encoding Functions

URL Encoding Reference

URL Encoding Reference

1
2
3

<html lang="en">
...
</html>

1
2
3

<html xmlns="http://www.w3.org/1999/xhtml" lang="en" xml:lang="en">
...
</html>

ISO 639-1 Language Codes

table-layout: auto | fixed

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" "http://www.w3.org/TR/html4/strict.dtd">
<html>
 <head>
  <meta http-equiv="Content-Type" content="text/html; charset=windows-1251">
  <title>table-layout</title>
  <style type="text/css">
   TABLE {
    table-layout: fixed; /* Fixed cell width */
   }
  </style>
 </head>
 <body>
 
  <table width="100%" cellspacing="0" border="1">
   <col width="160">
   <col span="9" width="60">
   <tr> 
    <td>&nbsp;</td>
    <td>1995</td><td>1996</td><td>1997</td><td>1998</td>
    <td>1999</td><td>2000</td><td>2001</td><td>2002</td>
    <td>2003</td>
   </tr>
   <tr> 
    <td>Oil</td><td>5</td><td>7</td><td>2</td><td>8</td>
    <td>3</td><td>34</td><td>62</td><td>74</td><td>57</td>
   </tr>
   <tr>
    <td>Gold</td><td>3</td> <td>6</td><td>4</td><td>6</td>
    <td>4</td><td>69</td><td>72</td><td>56</td><td>47</td>
   </tr>
   <tr>
    <td>Wood</td><td>5</td><td>8</td><td>3</td><td>4</td>
    <td>7</td><td>73</td><td>79</td><td>34</td><td>86</td>
   </tr>
  </table> 
 
 </body>
</html>

]]> http://phpforms.net/tutorial/css-reference/table-layout.html/feed 0 http://phpforms.net/tutorial/css-reference/right.html http://phpforms.net/tutorial/css-reference/right.html#comments Wed, 14 Apr 2010 07:51:38 +0000 bekhter_aa http://phpforms.net/tutorial/?p=3419

Browser	Internet Explorer				Netscape		Opera				Safari			Firefox
Version	5.5	6.0	7.0	8.0	8.0	9.0	7.0	8.0	9.2	9.5	1.3	2.0	3.1	1.5	2.0	3.0
Supported	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes

Short information

CSS	CSS2
Default value	auto
Inherited	No
Applied	To all elements
HTML analog	No
Reference to specification	http://www.w3.org/TR/CSS21/visuren.html#propdef-right

Description
This parameter defines distance from a right border of a parental element to a right border of an affiliated element not including a space, a field and width of a framework (fig. 1). Readout of coordinates depends on parameter position value. If its argument is equal absolute, in this case the parent is the browser window. An element position is defined from right border. In case of value relative the argument right is counted from a right border of an element starting position.

Syntax

right: value | percent | auto

Arguments
Any units of length accepted in CSS are applied as values, for example pixels (px), inches (in), points (pt), etc. Value of parameter right can be negative. In this case superpositions of different elements against each other are possible. If you specify the value in percent, the element position will be calculated depending on parental element width. The argument auto does not change the element position.

Example

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" "http://www.w3.org/TR/html4/strict.dtd"> <html> <head> <meta http-equiv="Content-Type" content="text/html; charset=windows-1251"> <title>right</title> <style type="text/css"> #leftcol { position: absolute; /* Absolute positioning */ top: 20px; /* Location from the top edge */ left: 20px; /* Location from the left edge */ width: 100px; /* Block width */ background: #fc3; /* Background color */ border: 1px solid #000; /* Frame parameters */ padding: 10px; /* Fields around the text */ } #centercol { position: relative; /* Relative positioning */ background: maroon; /* Background color */ padding: 10px; /* Fields around the text */ color: white; /* Text color */ margin: 13px 250px 0 130px; /* Indents around the block */ } #rightcol { position: absolute; /* Absolute positioning */ top: 20px; /* Location from the top edge */ right: 20px; /* Location from the right edge */ width: 200px; /* Block width */ background: #ccc; /* Background color */ border: 1px solid black; /* Frame parameters */ padding: 10px; /* Fields around the text */ } </style> </head> <body>   <div id="leftcol">leftcol</div> <div id="centercol">centercol</div> <div id="rightcol">rightcol</div>   </body> </html>

The result of this example is shown on the figure 2

Object model

[window.]document.getElementById("elementID").style.right

Cloud Computing

Cloud Backup Services

Cloud Backup

Backup Server