Enable direct access to the DRS array

if there is fabric connectivity between the sites, nothing is stopping you from doing that, zone the servers and you should be fine. Of course performance will depend on latency between the two sites.

Ok but i have few Q's;

- is it possible to mask a device from a remote array ?

- the devices are in ASYNC mode, then i need to pause the replica first and try to access the device or to type a script to pause all replication first?

- If i have two clariions then this should be the same case ?

Thanks

VSANs merged between site A and site B or you are using IVR with transit VSAN ?

yes , dedicated ports for the SRDF for the SYM case.. 

ok, so do you have dedicated VSAN for your SRDF ports or they are in the same VSAN as your hosts ?

How are the two arrays connected ? FCIP / DWDM ?

no IVR were used..

dedicated ports in dedicated "replication" VSAN ? Is this the VSAN that is trunked over FCIP link ? The reason i am asking is if you have dedicated VSAN for replication and dedicated VSAN for host connectivity, you will need to verify that your IVR topoligy includes the VSAN with hosts and storage in it.

yes its the one trunked over thr FCIP, IVR .. can you explain it in breif, AFAIK its a routing over VSANs ? i dont think its enabled in my case,,

Let's look at this sample topology, this is fabric A ...you would have similar setup on fabric B

diagram 2

so i have two sites that are connected via FCIP. Let's say top of the diagram is production and bottom of the diagram is DR site. In diagram one if you wanted to connect SRDF ports on array at production site to SRDF ports on array at DR site, you would have to merge VSAN 20. On your trunk you would have to allow VSAN 20. So this would work just fine for Replication VSAN but let say now you want to connect hosts at production site to array at DR site. In order to do that in digram 1, you would have to trunk allow VSAN 10 and merge both VSANs. This is where it gets dicey, you now have one big fabric that spans multiple sites. If your FCIP links starts bouncing, it may cause issues at both sites, production and DR (fabric will constantly merge and segment). So in order to prevent something like that i introduce you diagram 2 .  Here we are going to use IVR to interconnect only the devices that need to talk to each other. As you can see i have different VSAN numbers at production and DR sites. I also have a transit VSAN 30 that exists on both switches. Think of VSAN 30 as the bridge between the two sites. Only VSAN 30 will be trunked over FCIP link. When you define IVR topology you specify which VSANs can talk to each other. In this example we would create a topology where VSAN 10, 20 , 30 , 110, 200 could all talk to each other (without merging). This would allow you to IVR zone host at production site to array at DR site without merging VSANs. Once zoning is in place you would do regular masking as if that host was directly connected to DR array.  This topic most probably belongs in Infrastructure forum ..hopefully i did not confuse you

Thanks for the great efforts, but in my case i can add the host and storage vsan to the tunnel then it will be merged and i can do the zonnin the same way right?

you can but think about all the requirements and outcome if you merge two production fabrics. Do you have enterprise license on your switches ?

No , no enterprise license why you are asking ?

you need enterprise license for IVR. Do you have available ports on your DMX at DR site ? I was thinking before you merge your production VSANs, setup a test VSAN at both sites, place a test host at production site and place a DMX port at DR site and merge these two VSANs.  You need to verify that FCIP link performance is even sufficient to do what you are trying to do, many application are very sensitive to latency so if you link is not good enough, there is no point to mess with production VSANs. That's my opinion