Cluster Size Maximum?

[kornax]

I can't seem to find any documentation to show what the maximum amount of P4500 units should be in a cluster.

I currently have a cluster of 6 units. I have purchased more units, to support our growing environment.

My concern is, the size of the cluster is rather large, adding 2 more units will add 33% more space. Not only that, but the cluster houses a lot of data. Exchange/Oracle/VM's/User data etc. Adding these units will cause a restripe that will likely take many days, and all of these things will feel the performance impact.

Does anyone have any idea what the best practice is? Where do you get to the point that a cluster should not get larger? Any docs I am missing in my searches?

Any information would be great.

Thanks!

[Jitun]

Here are the Maxims found in Admin Guide and also Configuration Summary

Storage systems in the management group
The optimum number of storage systems in a management group ranges up to 20.
If the management group contains 21 to 30 storage systems, the Configuration Summary appears orange for that lineof the management group.
Exceeding 30 storage systems triggers a warning by turning that line red.
As soon as the total number of storage systems reduces below the boundary, the summary bar returns to the previous indicator, either orange or green.

Storage systems in the cluster
The optimum number of storage systems in a cluster ranges up to 10.
If the cluster contains 11 to 16 storage systems, the Configuration Summary appears orange for that line of the management group.
Exceeding 16 storage systems in a cluster triggers a warning by turning that line red.
As soon as the total number of storage systems reduces below the boundary, the summary bar returns to the previous indicator, either orange or green.

You can always Change The Local bandwidth priority
1. In the navigation window, select a management group and log in.
2. Click Management Group Tasks on the Details tab, and select Edit Management Group.
3. Change the local bandwidth priority using the slider.

A default setting of 4, at the Application Access end of the slider, is more appropriate for everyday situations where many servers are busy with the volume.
A setting of 40, at the Data Rebuild end of the slider, is most commonly used for quick data migration or copies when rebuilding or moving damaged volumes.

4. Click OK.

The new rate appears on the Details tab in the management group tab window.

This way you can Provide Application Access during business hours and give Stripping Priority during after hours.

Else to avoid the Re-strip completely (if you are not moving any volumes to the new Cluster) you can very well create a new Cluster and add the units there.

If you need to Move some of the volumes to the new Cluster, you can start that on weekend and make the Restrip priority High.
Generally I would expect the data transfer to finish by Monday.