RecoverPoint is one of those technologies that is a constant. The product has its own evolution and from where I am sitting today, the announcement of RecoverPoint for VMs at VMworld furthers the underlying fundamentals of the replication product WAY into the NOW and for the future. RecoverPoint is becoming ubiquitous for storage and application protection.
For those who are not familiar with the RecoverPoint infrastructure there are (3) basic components and I talk about them as they relate to the Virtual implementation vs. the physically oriented implementation using array families and EMC VPLEX.
1. The RecoverPoint Splitter – assures that copies of writes being sent to protected volumes are tracked and sent to the RecoverPoint appliance, this is installed on recommended each ESXi server in the clustered environment (same technology drives array based and VPLEX solutions)
2. The RecoverPoint virtual appliance – this is delivered in OVA format and customers can have 2-8 virtual appliances per cluster and provides the pathway for writes to be managed locally or remotely for continuous replication. (appliance based RPA is also part of physical solutions) The virtual appliance cluster can reside either on the same ESX as the protected VMs, on different ESX host or even a different ESX cluster.
3. RecoverPoint for VMs Plugin for vSphere Web Client – enables the management of RecoverPoint from vSphere UI and supported on 5.1 and 5.5 vSphere.
Now if you haven’t seen all of the listed benefits of using RecoverPoint for VMs in this architecture aside from being designed for VMware is that it enables protection at a “per VM granular” scale. Administrators can also replicate VMDKs or RDMs, test copies and promote failovers taking replication out of the storage layer and up to the VM layer. Let’s not forget it is also storage agnostic.
Here is the workflow of IO as depicted in the diagram.
Step 2. The ESXi Splitter (RP) intercepts WRITE and sends it to the Virtual Appliance via a Software iSCSI adapter
Step 3. The Virtual Appliance acknowledges the WRITE and in turn
Step 4., the ESXi Splitter sends the WRITE to the VMDK
5. The VMDK (via Storage) acknowledges the WRITE Splitter acknowledges the write to the VM
Looking at the various topologies as RecoverPoint for VMs we have the following:
1. Local Protection of VMs within the same ESXi cluster
3. Remote Protection of VMs to a remote site
In each RecoverPoint for VMs cluster, there has to be a minimum of 2 appliances for high availability. The cluster can scale up to a maximum of 8 appliances. A RecoverPoint for VMs system spanning two locations will have 2 RecoverPoint for VMs clusters – one at each location. RecoverPoint for VMs can scale out to include many independent clusters. In turn, each ESXi cluster can run multiple independent RecoverPoint for VMs clusters and as many as up to 32.
A RecoverPoint for VMs cluster can replicate many ESX clusters. This allows administrators the flexibility to locate the RecoverPoint for VMs appliances next to the protected workload (i.e. in the same ESX cluster), or alternatively – locate the RecoverPoint for VMs appliances in a separate set of ESXs on a separate cluster. All of the ESX clusters that run replicated workloads must run the ESX splitter on all nodes in that cluster and be able to communicate with the virtual appliance iSCSI interfaces. The virtual appliances themselves are required to have visibility to all the datastores across the ESX clusters where the protected VMs reside.
For additional flexibility, it is recommended that the splitter be installed on every ESXi server. In order for RecoverPoint for VMs to replicate, the splitter needs to be installed on the ESXs that run the source and the target replicated VMs. Since VMs may be moved around across ESXs, all the ESXs that are expected to run replicated VMs should have the splitter installed. Replicated VMs will not be able to be migrated to an ESX that does not have a splitter in it.
Overall, there are many more details that apply to the RecoverPoint for VMs solutions but this is a preview as to how lightweight of an implementation can be vs. how scalable and flexible protection can be in the customer environment.
And if you haven’t yet, be sure to watch our video demo of managing a RP for VMs solution with XtremeIO.