EMC VNX – MirrorView introduction

MirrorView offers block-based replication features for EMC VNX. There are two remote mirroring modes:

  • MirroView/S (Synchronous)
  • MirrorView/A (Asynchronous)

The MirrorView is a technology that mirrors and active block data set to a remote VNX system. It’s a great solution for a disaster recovery replication. MirrorView is LUN centric which means the replication is from a primary LUN (on our primary VNX) to a secondary remote LUN (typicaly our disaster recovery site).

MirrorView/S

MirrorView/S provides synchronous replication over a short distances.  Since it’s synchronous solution, the RPO (Recovery Point Objective) is zero. The data flow of MirrorView/S is:

  1. Host attached to the primary VNX system initiates a write;
  2. The primary VNX replicates the data to the secondary VNX system;
  3. The secondary VNX acknowledges the write as complete back to the primary VNX;
  4. The primary VNX acknowledges the write as complete back to the host.

It is important to understand the data flow of MirrorView/S. As a consequence the RTT (Round Trip Time) between the two VNX system should be less (or equeal) than 10ms. If the RTT is high, then the host will higher response time, because it will take longer time to acknowledge the write request.

MirrorView/A

MirrorView/A provides replication over long distances. It can be used for replication between VNX arrays, where RTT (Round Trip Time) is high, but should not be higher then 200ms. MirrorView Asynchronous works on a periodic update model that tracks changes on the primary side, and then applies those changes to the secondary at a user-determined RPO (Recovery Point Objective) interval.

With MirrorView/A replications, writes are acknowledged back to the host once the primary VNX receives them, which basically gives no impact on the production environment (whereas with MirrorView/S all writes has to be acknowledged by primary and secondary VNX system).

The dataflow of MirrorView/A is:

  1. Host attached to the primary VNX system initiates a write;
  2. Primary VNX system sends and acknowledge to the host;
  3. Primary VNX system tracks the changes and replicates the data to the secondary VNX system using the user-defined RPO;
  4. Secondary VNX system receives the data and sends an ack back to the primary VNX.

Consistency groups

Both MirrorView/S and MirrorView/A supports consistency groups. Consistency groups are used when a set of LUNs require a consistent write ordering. For example, if you have a VMware datastore build upon 6 LUNs, it would require for all 6 of them to be in a consistent state in relation to each other for proper functionality.

 

2 thoughts on “EMC VNX – MirrorView introduction

Leave a Reply

Your email address will not be published. Required fields are marked *