By Jim Damoulakis, ComputerWorld.com,
As more organizations adopt replication as a primary component of disaster recovery, it's important to better understand some of the variances among replication technology and to clearly set expectations with application owners when planning replication deployments.
A common area of confusion in dealing with replication is the distinction between consistency and synchronicity. Many newcomers to replication tend to focus on synchronization issues when, from a recovery perspective, consistency may be the true requirement from an application perspective.
So what is the difference and why is it important? Synchronization implies complete and continuous fidelity between local and replicated data stores. With true synchronous replication, a write operation is not acknowledged until it has been written to the local storage system and replicated to the remote storage system. This certainly provides a very high degree of consistency, but it also carries with it high costs and significant limitations regarding distance and latency that can impact application performance.
Synchronous replication is found primarily in the domain of the top tier of enterprise storage offerings and is usually reserved for those applications that are characterized by very high transaction rates where the recovery and re-execution of lost transactions would be difficult and costly.
The majority of replication is therefore of the asynchronous variety - meaning that there is some degree of variance, based on change rate and available bandwidth, between the local and the replicated targets. In other words, by definition, the source and target are inconsistent with one another.
However, consistency still plays a critical role in the recoverability of asynchronously replicated data. The key is in understanding the interdependencies among related data components of a particular business function and ensuring that they are consistent among themselves at any given point in time at the target location. They may lag behind the original, but as long as they are equally behind, the function or application should be recoverable.
While the notion of consistency groups is well established among enterprise-class storage systems, it may be less so for other forms of replication. Understanding consistency requirements and the ability of replication technologies to meet them should be a high priority consideration in disaster recovery design.
Jim Damoulakis is chief technology officer of GlassHouse Technologies Inc. , a leading provider of independent storage services. He can be reached at jimd@glasshouse.com.
Copyright © 2008 IDG. All rights reserved.