I have two data centers on our campus. The first is our primary data center and the second is used as a business contiuity site. They both have a Clariion CX3-10's, 2 large servers running VMware. We are using Mirrorview to replicate from one SAN to the other. What I am looking for is there any way to load balance the two SAN's? Right now, if we have a major disaster with our SAN in our primary data center, we have to go over the our BC site, promote the second SAN as a Read/Write, associate each VMWare host to the LUN where their data is stored to get them running. With 50 servers, this will take a long time. I am looking for a solution that we can use both SAN simultainous so we can cluster all 4 VMware boxes so if the primary SAN goes down, we automatically run off the one in the BC site.
That's an engaging question. There are a couple of ways to tackle this based on the key driver of your organization's RPO/RTO tolerances and goals. The first part of your question was "...is there any way to cluster all four servers across two SANs..." In order to do that, you need some level of storage virtualization. You could look into the storage Vmotion capabilities of VMware so that the vmdk's float back and forth between VMFS volumes on either SAN. This can ride on Mirrorview. But as you've noted, it goes one way. You could also use onboard CX Recover Point splitters to accomplish dual writes simultaneously across your fabrics to each SAN. VMware Site Recovery Manager rides on this quite nicely.
Another approach might be to employ Invista to completely virtualize the entire SAN infrastructure so that the ESX hosts don't know which SAN they are talking to, and don't care. VMFS volumes are available for writes as needed.
Needless to say, there are some cost factors involved here - so a conversation about RPO/RTO at the business level is usually a very good idea.