Review - Disaster Recovery using vSphere replication and SRM
This new book from Packt shows just how simple SRM and vSphere replication can be to setup. The book covers the original SRM package, then vSphere replication , then using both together.
The first section of the book covers SRM and explains that it is mostly an orchestration tool and doesn't really do the actual work on it's own. With SRM you create protection groups on Site A and they are replicated to Site B. Part of the complexity is that SRM typically works with SAN based replication using SRA's plugins, so the unit of replication is the SAN volume, not the VM. The minimal configuration of SRM is discussed as needing SRM installed in both sites, a storage array with SRA at both sites and vCenter at both sites. SRM can manage the mapping of resource pools at Site A to Site B, and also the mapping of network port groups.
Because SRM replicates volumes using array based integration tools, there may be a need to re-deploy VM's in a more organized fashion to the volumes aligned with the protection groups you have designed.
This section also covers how SRM is designed to be tested and perform actual fail-overs without losing data. Before a planned Migration steps are taken to shutdown Site-A and perform one last sync to make sure all changes get propagated. SRM can manage IP changes to conform to Site B as well as startup order for multi-tier applications.
Next the book covers vSphere replication which works across any type of vSphere compatible storage. It can replicate from one type of storage to another, even from NFS to iSCSI devices. It is included in entry level Enterprise Plus and higher offerings makeing vSphere replication very attractive for entry level customers. vSphere replication , however, does not include orchestration for fail-over or fail-back, this must be done manually.
This section also explains that vSphere replication does not require shared storage or a vCenter at the recovery site, only a single VM recovery appliance, making this ideal for keeping costs low at the recovery site, and moving vCenter and other licences over after a DR event.
The final section covers using vSphere replication along with SRM. This of course requires a vCenter at both sites, however the storage need not be supported by a SRA providing a lot more flexibility in storage selection. By using SRM with vSphere replication all the orchestration advantages of automated fail-over and fail-back are now available.
While the book has lots of step-by-step instructions for installing and using these products, the sections on high-level architecture are only small fragments thrown in-between the major howto sections. It would have been nice to see an initial chapter devoted to architecture and planning and your options for a minimal or full-up DR site.
Overall this book presents a lot of practical knowledge in setting up a DR site and the differences and consequences of using SRM, vSphere Replication or both together.