Automated disaster recovery (DR), shortened downtimes and simplified business continuity are just some of the benefits that VMware Site Recovery Manager (SRM) offers. Controlled by VMware VirtualCenter and working in conjunction with new storage replication adapters on storage systems, Site Recovery Manager stands poised to revolutionize how companies think about and manage DR and data backup for their VMware virtual server environments. Yet for companies that lack a carefully planned and documented DR plan for their virtual servers, implementing Site Recovery Manager will remain a difficult, if not impossible, task.
There are five main components that must be in place prior to implementing VMware SRM in an environment:
- VirtualCenter Server 2.5
- Virtual Infrastructure (VI) Client 2.5
- Oracle or Microsoft SQL Server database server
- Storage systems with a storage replication adaptor (SRA) that appears on VMware's hardware compatibility list
- VMware ESX Server 3.0.2 Update 1 or ESX 3.5 Update 1
But even with these building blocks in place, VMware Site Recovery Manager is still far from a turnkey operation. Chris McCall, director of product marketing for LeftHand Networks Inc. says that LeftHand has successfully tested Site Recovery Manager in its own labs and demonstrated it at events like VMworld Europe. However, LeftHand has no customers currently use it in production (though it was just released on June 22), partly due to their need to first develop a DR plan and then bring in the appropriate hardware and software that Site Recovery Manager requires to support it.
McCall recommends that companies have a completely thought out DR plan prior to bringing Site Recovery Manager in house as users can't take any component that makes up their Site Recovery Manager configuration for granted. For instance, the storage systems that are used in conjunction with VMware Site Recovery Manager need to have the following characteristics: the same vendor's model at the production and DR site; replication that is integrated and certified with SRM; and a storage replication adaptor that is supported by Site Recovery Manager. The storage replication adaptor allows Site Recovery Manager to communicate with the storage system and manage the replication of data of specific storage system LUNs that Site Recovery Manager needs for the automated recovery at the second site.
Though the need for the storage system to appear on the VMware HCL isn't explicitly stated as a prerequisite for SRM, companies should view it as one. LHN has found that as many as three-quarters of companies won't implement storage hardware unless it appears on VMware's hardware compatibility list, which was a problem for LeftHand Networks two or three years ago.
Companies should also not assume that VMware Site Recovery Manager automatically recovers applications on the virtual machines on ESX hosts. Site Recovery Manager only recovers the virtual machines hosted on each ESX server, not necessarily the application that the VM hosts. Unless the application happens to start as part of boot-up process of the VM on the ESX Server at the DR site, companies will still need to put in place separate processes to recover these applications.
Glen Rhodes, CA's vice president, product marketing for recovery management and data modeling, points out that VMware Site Recovery Manager is focused on infrastructure (servers and storage) -- not the application layer. So if failures occur at the application layer of a virtual machine while the infrastructure layers remains operational, no failover to a secondary site will occur because Site Recovery Manager doesn't detect the application failure. "Companies need application-aware software that automates the scripting and failover to bring the application back online on another server [physical or virtual] at the remote site," says Rhodes.
VMware Site Recovery Manager is a needed new feature for companies looking to automate disaster recoveries of their VMware virtual server environments. However, it's anything but a quick and simple configuration; it requires that companies first devote a substantial amount of time to planning and documenting what virtual machines they want to recover and in what order they wish to recover then. Only once that is done should companies contemplate implementing Site Recovery Manager in their environment.
About the author: Jerome M. Wendt is lead analyst and president of DCIG Inc.
Do you have comments on this tip? Let us know.
Please let others know how useful this tip was via the rating scale below. Do you know a helpful backup tip, timesaver or workaround? Email the editors to talk about writing for SearchDataBackup.com.