Failover between two Datacenters with Open-E and VMware Site Recovery Manager
We are looking to get a secondary data center location in a different state. The link between the primary and secondary site is a 100Mb connection with about 7ms latency between the two. This will be a private, dedicated link.
We currently have two Open-E boxes setup with failover/MPIO over block-IO iSCSI to a few VMware clusters.
Now, the question is:
1 - Do we take one of the Open-E boxes to the secondary site?
or
2 - Do we setup a new iSCSI cluster there and use VMware Site Recovery Manager (5) to send the data over the link?
Benefit of 1 = both boxes are in sync. Downside of 1 = the I/O will slow down even more since it's only as fast as the replication link. Right now, they are connected with 10Gb between the two (though we noticed it rarely even hits the 100Mb mark in the graphs, so something is definitely wrong since we're seeing major I/O latency, and the drive RW tests from the console reveal that they are capable of handling the 10Gb link and more).
Benefit of 2 = sync doesn't affect the latency of the local Open-E cluster and is separate. Downside of 2 = since there's no SRM tie-in with Open-E, the data sync interval for SRM can only come down to every 15 minutes.
So, are there any others out there doing something similar? SRM with Open-E, or a long-link replication with iSCSI?
With the 10Gb sync as slow as it is already, it seems simply moving one of the boxes over to the secondary site would be the easiest solution?
Any advice?
MJP Technologies - Intel Technology Provider Platinum Member