HP Continuous Access Behaviour

[IvanForceville]

We're currently doing some testing with CA between two EVA's with an ISL of 50MBit.

We created a 50GB Vdisk with one Virtual machine of 10GB on it.

We started replication from one site to the other which took about 2-3 hours.

When we failover to the other side, HP CA start sycning the other way and replicates the entire Vdisk to the original side, knowing that the data is 99% the same. So, every time we failover/failback the Vdisk is replicated entirely (we presume)

Is this normal behaviour? I thought that HP CA would Check what's left of the LUN on the source side and only move the delta's?

?

[Uwe Zessin]

That does not sound correct. Can you give a little more technical details about your EVA configuration ? (firmware revision, using sync/async replication, ...)

[IvanForceville]

We have two EVA4400's with FW 09522xxx and are using Async replication.

Distance between sites is 50-60km.

When we failover to the other side, it takes about 1,5 hours to synchronize towards the original side, knowing that there is only 1 Virtual Machine on the LUN which hasn't changed.

Tested this a few times today.

[Uwe Zessin]

Do you switch to sync mode before the failover?

[IvanForceville]

Not realy... But honestly, I need to check. It appears that the copy is in Sync mode but we definately don't explicitly set it.

[Víctor Cespón · Accepted solution]

Looks like you're doing an UNPLANNED failover, so a full copy is done.

Read the RSM user guide, it's the best way to understand Continuous Access.

http://bizsupport2.austin.hp.com/bc/docs/support/SupportManual/c02012745/c02012745.pdf

Look for "PLANNED failover"

[IvanForceville]

Guys,

You we're correct. There is a difference between Planned and Unplanned failover... Since we were using Vmware Site Recovery Manager to do the failover the process was automated... didn't catch it.

Thx... close case.