Unsolved
This post is more than 5 years old
78 Posts
0
1592
May 25th, 2010 05:00
Replistor performance
I am using RepliStor 6.4 HF1 on all my servers. I am trying to replicate 15 VM servers to 3 VM servers setup in an image collector configuration. The servers I am trying to replicate range from files share servers, SQL servers and exchange servers. I am running this across a point to point 10MB VPN. The issue I have is I can't run RepliStor fast enough to prevent an enormous amount of Kernel Log files from accumulating on some of my servers which in turn fills disk space. I have adjusted the performance setting from WAN to Fast LAN trying to find a good setting for performance. How can I increase my Forwarding Ops/sec on my servers to prevent log build up issues? Am I asking to much of RepliStor with this many servers across a the VPN?
jschrade1
2 Intern
•
106 Posts
0
May 25th, 2010 06:00
Indyemcuser,
Without more information it is difficult to make general recommendations. However, host based replication of servers vary based on the speed of the connections, the latency of the connections, the overall rate of data change and the performance of each individual machine in the group.
RepliStor does have Many-to-One capability but there is no formula for determining when you have exceeded the ability of a single server to accept the data changes from more than one Source node. We have seen customers replicate data from 15 or more Source nodes to a single Target node. We have also seen situations where one Source node cannot send the data fast enough to overload a single Target node.
There are some Performance Counters that can help you determine whether a Target node is overloaded (Target Queue Count) but that can only tell you after the Target node has reached its limit and cannot keep up. At this point you should probably start by adding another Target server. However, remember that the Disk I/O may still be going to the same set of disks and/or HBA. If the physical disks cannot keep up with the load and the disk queues are already too high, then adding another Target RepliStor node will only increase the load on the disk subsystem.
I would also recommend you open a case with EMC Support regarding any more specific recommendations.
Thanks…JS
indyemcuser
78 Posts
0
May 25th, 2010 06:00
To add, is it possible to redirect the Kernel Log files (Data Directory) to a drive other than the OS dirve? Each of my replicating servers must run with the Boot Loader Driver enabled since these are source to target in an image collector configuration.
indyemcuser
78 Posts
0
May 25th, 2010 06:00
Thanks for the reply. For what ever it's worth, I am running 3 image collectors.
A support incident might be in order.
jschrade1
2 Intern
•
106 Posts
0
May 25th, 2010 08:00
Under other circumstances you can use a different drive for the Data Directory. In this configuration, where you are replicating the entire C: drive, it must stay on the C: Drive. However, you can increase the size of the C: drive to accommodate the Data Directory. You would need to refer to the VMware documentation on how to expand the C: drive.