Is there anything on the roadmap to allow for multiple XtremIO arrays managed by a single XMS management server to retain all of their individual array limits rather than consolidating to the single XMS limits?
Right now, it seems/states that when you manage multiple XtremIO arrays with the same XMS server, the scalability limits become global across all arrays managed by that XMS, so if I have 2 arrays managed by that XMS, I will still only ever be able to create, let's say, 8192 volumes total across all those arrays that are being managed by the single XMS. Whereas if I stick with a 1:1 XMS-to-XtremIO array management model, each array would have the limit applied to them of 8192 each.
Thanks for the feedback Keith. We are looking at improving those scalability numbers in multiple ways. You will some of these numbers changing in our future releases. I don’t have a committed date for that change at this point of time.
Thanks Avi...I will try to stay close to my account team as well. With one of my customers, I think I am going to break it back down to a 1:1 mapping of XMS to XtremIO array for the time being since RP replication is in place and I have concerns around the snap counts being global.
dynamox - thats correct. The maximum number of LUNs (8192 at this time) is defined at the XMS level. So, if you have two separate clusters being managed by the same XMS, you can have a maximum of 8192 LUNs across those two clusters.
Oh boy does XtremIO have a long way to go to catch up to your other element managers (Unisphere for VMAX and even VNX Unisphere). Let's assume someone did not know about this limitation, can they break off one cluster from an existing XMS and have it managed by another XMS ?
I have to do just this now given the VIAS limitations and after reviewing with EMC and having done something similar for a customer that lost their XMS, the answer is yes and the general process (to be performed by EMC or authorized personnel) is this:
Deploy a new 4.x OVA
Upgrade the OVA to the cluster version of the array you will be adding to it so they match (EMC/Partner does this)
Remove the cluster from the existing multi-cluster vXMS
4. Add the cluster to the new vXMS
I'm sure others will chime in from EMC if I am off base.
Re: “Let's assume someone did not know about this limitation, can they break off one cluster from an existing XMS and have it managed by another XMS?”.
Answer – yes. You can do that. You can always move the XtremIO from one XMS to the other. There are CLI commands (remove-cluster and add-cluster) that does exactly that.
Thanks for the feedback Keith. We are looking at improving those scalability numbers in multiple ways. You will some of these numbers changing in our future releases. I don’t have a committed date for that change at this point of time.
Just wanted to note, that 2 years after this comment, these limits are still in place, as per Release Notes for 4.0.25-27 (P/N 302-004-060, REV 02, October 17, 2017). Per those release notes, the scalability numbers of the XMS have not changed since at least 4.0.0-64, circa June 2015 (that was the earliest release note available in SolVe Desktop for XtremIO 4.0, which was when multiple clusters per XMS was introduced).
Kumar_A
2 Intern
•
727 Posts
4986
0
Posted December 15th, 2015 07:00
Thanks for the feedback Keith. We are looking at improving those scalability numbers in multiple ways. You will some of these numbers changing in our future releases. I don’t have a committed date for that change at this point of time.