Array based replication is the ability to have data automatically copied to a remote site creating the foundation of a solid disaster recovery strategy and as a result many storage arrays have a replication option. In fact many vendors provide replication at no charge. Their motivation is that a replication option is going to require that you buy more storage hardware, probably from them.
Replication is the ability to copy data from one location to another. Most replication products have the ability to perform these copies at a sub-file (block or byte) level so that only the smallest chunks of data need to be copied to another site. This allows for optimal use of bandwidth. Having the storage system be responsible for the movement of data to a disaster recovery site seems like an obvious solution to a nagging problem; how to seed and keep a remote DR site up to date. Array based replication though has some downsides that storage managers should consider prior to investing in them, even if the capability comes free with the storage array.
The Downsides to Array Based Replication
The first problem with array based replication, in almost all cases, is that it requires a near identical unit from the same manufacturer at the DR site. The DR site though may not need the same level of performance or redundancy as the primary, and a less expensive system may be a more appropriate fit. Software based replication tools like those from Vision Solutions enable replication between storage arrays of different vendors.
Another problem that can occur is when there are multiple storage systems in the data center. Each storage system requires a similar storage system from the same manufacturer as a replication target. This use of multiple storage systems results in inefficiencies around managing these arrays because management of the replication process cannot be done from a single interface. Each array will have its own software that needs to be configured and monitored. This increases costs significantly. Once again software based replication allows any storage system to replicate to any other. It can also support a many to one replication strategy which allows the DR sites storage to be consolidated into a single system.
This inability to mix storage hardware also means that adding new hardware from a new vendor must be considered carefully, since the introduction of a new vendor will mean learning another replication module and buying a second storage system for the DR site. It may also mean that a newer more innovative storage vendor may not be considered because they lack a replication capability. Replication is often one of the last capabilities that a new vendor adds to their storage system. However, a software based replication tool frees you to add new vendors as it makes sense.
The third problem is that array based replication software is not application aware. These modules just replicate blocks of data as it changes on the array with no understanding of the application that is changing that data or how other servers may be related to that application. Software based replication software that is application aware can ensure that clean copies of data are captured. They can also make sure that groups of servers that are dependent on each other can be kept in sync. Why or how?
Array based replication seems like an ideal option for businesses looking to establish a DR site, especially if the capability comes with replication for “free”. The problem is that free isn’t really free. A good storage manager factors in the cost to support multiple replication processes from each vendor and the extra cost to power and cool the additional storage systems in the DR site. These costs should then be compared to storage independent management and replication software to determine the best solution for array based storage protection.