How to validate that the VRA has sufficient resources for replication
Viewed 2553 times
- Open the ZVM UI.
- Open the VPGs tab.
- Select the relevant VPG.
- Click on the “Protected VMs” tab.
- Expand the ‘Protection Host’ and ‘Recovery Host’ tabs.
- The Local memory refers to the local buffer – the buffer that is involved when the VRA acts as a production VRA, collecting and sending I/Os to the peer VRA.
- The Remote memory graph refers to the buffer that is involved when the VRA acts as a recovery VRA.
- Connect to the vCenter UI (Both Flash and HTML 5 are fine).
- Click on the relevant VRA -> Monitor -> Performance -> Advanced.
- Click on ‘Chart Options’.
- We will review three different charts Datastore, CPU and Memory.
This check should be performed on the recovery site.
Under the 'Chart Options', select Datastore and make sure to check both Read Latency and Write Latency and click OK.
By default, Zerto will start throttling I/Os if either one of the latency values goes above 40 ms. This was designed to protect the actual storage devices from overheating.
This value is tweakable, but Zerto strongly recommends approaching this matter from the storage perspective to ensure the longevity of your storage devices.
If you’d like to tweak the bad I/O throttling value, please contact Zerto support.
- The ‘Consumed’ graph will show us the consumed memory from the Host (the ‘total’ as far as the VRA is concerned).
- The Active graph will show us how much is being utilized at any given point in time.
- The ‘Host consumed %’ will translate the ‘Active’ graph to a percentage out of the ‘Consumed’ value.
If the usage is above 70% Zerto recommends trying to balance the load across more VRAs or add more CPU to the VRA.