RecoverPoint for VMs Architecture Part I

when planning a disaster recovery solution, there are various factors that need to taken into account. factors like RPO and RTO as well as vendor-agnosticism are those, to name a few.

RP4VMs is one of the solutions specifically made for VMware environment by DELL EMC, which gives you the ability to Point-in-Time recovery your virtual machines. This is a perfect options, particularly when it comes to disasters that involves data corruption or human errors and not complete data loss. The other benefit of using RP4VMs is that, it doesn’t matter what underlying storage you are using. It supports variety of protocols: SAN, DAS, NAS, iSCSI, FC, FCoE, vSAN, SIO and in contrast to Classic RecoverPoint which replicates and protects the whole LUN sitting on certain storage systems, here in RP4VMs you can replicate on the basis of individual VMs.

RecoverPoint Architecture

RecoverPoint uses a journal based replication to protect virtual machines. You need to consider a journal volume which stores all changes, and this will be used to go back to specific point in time. The concept of write splitter is used here, which in VMware environment sits in vSCSI layer, the splitter VIB is installed on ESXi when deploying the first vRPA (Virtual RecoverPoint Appliance).

IO Flow

when a virtual machines issues a write command to the VMDK, the following occurs:

  1. VM sends a write IO to VMDK.
  2. The ESXi Splitter intercepts write and sends to the vRPA protecting the VM.
  3. vRPA acknowledges the write and replicates it.
  4. ESXi Splitter sends write to VMDK.
  5. VMDK acknowledges the write.
  6. Splitter acknowledges the write to the VM.

In the next blog post, I’ll go through more details.

Leave a Reply

Your email address will not be published.