VMware bakes disaster recovery into vSphere stack
NFS spice added
VMware has rounded out its vSphere stack of software around the ESX Server 4.0 hypervisor, with integration of vCenter Site Recovery Manager that started shipping in June.
While VMware likes to make it look like the vSphere stack consists of lots of different code, thereby making it worth lots and lots of money, there are really only two pieces to the VMware stack: the hypervisor - ESX Server - and the management tools for it - vCenter.
Most of the wonderful gadgetry that VMware talks about - with vSphere, it is vSwitch, vSecurity, vCompute, and a bunch of features within each set- is in the hypervisor itself. (Psst. Don't tell Wall Street.)
Site Recovery Manager, or SRM for short, sits on the vCenter side of the VMware duopoly, and it allows for the automatic failover and recovery of virtual machines and their applications on a network of machines running the ESX Server hypervisor.
With vCenter SRM 4.0, the ESX Server 4.0 hypervisor can now have its VMs managed and failed over by SRM. Up until now, SRM only supported earlier ESX Server releases.
The latest SRM 4.0 update also adds support for the Network File System (NFS) protocol to link the ESX Server instances out to storage and perform the failover of VMs. Prior SRM releases had to talk to storage through the iSCSI or Fibre Channel SAN protocols. VMware says that it now can replicate data and provide disaster recovery for servers using disk arrays from a dozen different vendors.
SRM 4.0 also has a new feature that allows for a single shared recovery site to be the target for failovers for a whole slew of production sites running ESX Server. This many-to-one failover support will make disaster recovery a whole lot less expensive, as long as the disaster in question doesn't take out a lot of virtual machines at the same time.
VMware's parent company, disk array maker EMC, was first in line to crow about its support of SRM 4.0 on its various storage products. EMC says it has created storage replication adapters for SRM 4.0 that allows it to reach into Symmetrix Remote Data Facility (for high-end Symmetrix arrays), Celerra Replication (for servers linked to the array through NFS and iSCSI ), and MirrorView (for Clariion) and RecoveryPoint (for geographically distributed storage arrays).
RecoveryPoint allows for synchronous and asynchronous replication between storage arrays that are hooked to the source and target server platforms on which ESX Server are running.
SRM 4.0 is priced $1,750 per processor where vCenter runs excluding support, and it requires the vCenter Server 4.0 management tool that is part of the vSphere 4.0 stack. The tool can manage VMs running atop ESX Server 3.0, 3.5, or 4.0, so you don't have to upgrade all of the hypervisors at once just to get the better SRM features. The software also requires that some sort of supported data replication between servers and their storage be in place.
SRM was launched in September 2007 along with the embedded ESX Server 3i hypervisor, but didn't start shipping until May 2008. The product has over 2,000 customers to date, according to VMware. ®