vSphere Replication was introduced more than ten years ago. It has been continuously improved since then. I am pleased to share that its fundamental replication capability and architecture have been modified and enhanced to enable new capabilities immediately. And it opens the door for significant future improvements as well. Let’s dive in and learn more.
This initial availability (IA) announcement is specifically for vSphere Replication when used as part of VMware Site Recovery for VMware Cloud on AWS. For now, it requires VMware Site Recovery as both the source and target. We are planning to expand this in the future. Please see this blog for details about VMware’s IA/GA release system.
Enhanced replication for VMware Site Recovery
This enhanced replication for VMware Site Recovery can provide 1 min RPOs, down from 5 mins. It also supports auto-load balancing of replicated VMs as well as automatic scaling. It provides these capabilities while still supporting all the current scale and other capabilities of vSphere Replication.
To understand how it can provide these improvements, it is necessary to understand the changes made to the fundamental architecture of vSphere Replication. Let’s do that now.
Figure 1: Components and traffic flow of vSphere Replication
vSphere Replication currently routes traffic to the replication server component on one of the appliances at the target site. From the replication server, it utilizes NFC (Network File Copy) to write the replication data to the target storage through an ESXi host.
Figure 2: Components and traffic flow of enhanced replication for VMware Site Recovery
With enhanced replication for VMware Site Recovery, the replication traffic goes directly to the host at the target site, bypassing the replication appliance. This change, along with the addition of a connection broker component added to the vSphere Replication appliance, allows for backward compatibility while at the same time supporting a more optimized data path allowing for a reduced RPO of 1 minute, as well as increased simplicity, automatic scalability, and automatic load balancing.
These additional capabilities are provided while still supporting the same scale of 4,000 VMs per vCenter, as well as support for all the other features and capabilities of vSphere Replication like multiple point-in-time snapshots/rollback points, support for encrypted VMs, network compression of replication traffic, encryption of replication traffic and support for application quiescing for Windows VSS as well as current Linux operating systems.
Auto-Scaling and Automatic Load Balancing
With the addition of the connection broker and the change in replication traffic flow, we are now able to automatically monitor the number of replications being directed to each host and to reallocate or move replications as required to balance the number of replications between hosts as hosts are added and removed from the target site. This allows enhanced replication for VMware Site Recovery to support both automatic scale-up/scale-down and automatic load balancing of VM quantity across hosts at the target site.
Future posts will provide additional details about this exciting new evolution of vSphere Replication. We look forward to hearing how customers put these exciting capabilities to work to provide enhanced protection and easier management for their most important workloads.