Disaster Recovery Replication

Disaster Recovery Replication

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.