[GitHub] rafaelweingartner commented on issue #2500: Restrict the number of managed clustered file systems per compute cluster

[GitBox <gi...@apache.org>]

rafaelweingartner commented on issue #2500: Restrict the number of managed clustered file systems per compute cluster
URL: https://github.com/apache/cloudstack/pull/2500#issuecomment-377227548
 
 
   @mike-tutkowski, we have discussed this in an e-mail that is not public, but I think it is interesting to make it public domain. Maybe @Syed can add more information later here as well.
   
   When I read ” XenServer 6.5 can support around 500 - 600 SRs”, it feels that I will have a hard limit that might be complicated to manage in some deployments scenarios. This means that I might not even consider the technology because of this said limit. However, this 500-600 number is a not a limit per se.  You are considering it as a limit because of the time that the XenServer takes to create SRs. This means, for every SR that is being created, XenServer is taking a little longer to create the next one. The problem is probably due to some poor coding around UUID checkings and validations before creating some metadata information; the problem seems to grow linearly (according to what I saw in the data presented by you and Syed in a presentation).
   
   The point is that, even though it might take 150 seconds to create the SR number 1000, the performance of the SR is not affected (as far as we discussed).  Therefore, if the admin is ok with this, he/she can use a higher number of SRs. That is why you externalized this configuration. However, without making a difference between SR performance to write and read, and the bottleneck in the creation of the SR, we might give the idea that it is not supported more than this 600 number, which is not true. The 600 number came after some tests execute by Syed, and to create the SR 600 it was taking something like 80 seconds or so, and that was considered too much for the context he was working on.

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
users@infra.apache.org


With regards,
Apache Git Services