HPE EVA Storage
1748169 Members
4270 Online
108758 Solutions
New Discussion юеВ

Re: CA replication causing entire EVA slowdown

 
Jeremy C
Regular Advisor

CA replication causing entire EVA slowdown

Continuous Access replication between two EVA 8100's. Even after reducing replication to just a few vdisks (all on FC storage), the source EVA was having occasional fits of very high latency (approaching 1000ms) on two ports of controller A. I've since removed all DR Groups and everything is running smooth ( < 10ms latency).

I do not believe that the CA bandwidth is the issue and nothing was going into a logging or merging state.

Any ideas or things to check for?

10 REPLIES 10
Marcus Schack
Trusted Contributor

Re: CA replication causing entire EVA slowdown

Jeremy, how did you do your CA setup in regards to cabling and zoning? How far away is the second EVA? Here is a doc that might help.
V├нctor Cesp├│n
Honored Contributor

Re: CA replication causing entire EVA slowdown

Synchronous replication?

What's the distance and the delay on the replication link?

Replicating on FATA disks on destination EVA?

Those are the 3 main causes of what you see.
Jeremy C
Regular Advisor

Re: CA replication causing entire EVA slowdown

We have the two-fabric design as was the best practice, apparently up til Nov 2009.
- 2 san switches at each site
- 600Mbps fibre between sites
- all asynchronous replication
- Source vdisks, destination vdisks, and logs all on FC disks.
- distance is about 12 miles
- delay unknown

According to that Nov 2009 document, we should change to the six-fabric design. Was this problem a known reason why HP has recommended going to this six-fabric design? I'd like to know that this is the solution to the current problem before doing all that worked to change to six-fabric.
Marcus Schack
Trusted Contributor

Re: CA replication causing entire EVA slowdown

I would not change it yet.

You say you have 600Mbps fibre between sites. Is that peak or always on?

Also, do you have the proper SFPs in the ports of the switches to go the distance of 12 miles?

Is the port proper configured for that distance?
Jeremy C
Regular Advisor

Re: CA replication causing entire EVA slowdown

The 600Mbps is always available but typical usage has been about 50Mbps.
Yes, I believe the hardware and port config to be correct. It's been working great for several years. Only the last few days have there been problems.
Marcus Schack
Trusted Contributor

Re: CA replication causing entire EVA slowdown

Ok.. so what has changed in the last few days? Anything on your end? Check your EVA and switch error logs, just in case. Did they cut your bandwidth?
Jeremy C
Regular Advisor

Re: CA replication causing entire EVA slowdown

As far as I know, nothing has changed and I don't see anything in the switch or EVA controller event logs that jumps out at me.

If HP recommends the six-fabric config as a best practice then shouldn't I change to that regardless?

Jeremy C
Regular Advisor

Re: CA replication causing entire EVA slowdown

I just noticed that the Command View EVA clock was about 11 hours into the future for both EVAs. I've resync'd it with the management server. Could this have caused such problems?
Marcus Schack
Trusted Contributor

Re: CA replication causing entire EVA slowdown

No, I would not change just because Best Practices maybe have changed. You environment has been working for quit some time and only recently had an issue. I would be looking for what has changed/broken. Have you thought about logging a call with HP?