what Operating systems are currently running ?

AIX 5.3 and Windows 2003

is windows 2003 clustered ?

I have to HBA's attached to each hosts with alternate paths define for each SP. I have total of 8 HBA's on the CX700. I am running powerpath on both hosts. My thought was to utilize the additional four ports I currently have and configure them on the switch once that is accomplished switch on Storage processor at a time to the one attched via switch. I am assuming I have to change the config files on the server. Am I on the right track ?

** two ports on each host

** 0 and 1 for the AIX server

** 2 & 4 listed for the windows server

No its not

Do you have multiple HBAs in each host, and some kind of failover software running on each host as well?

If so this could be fairly simple (and no outage required). If not, then you will have to take an outage to insert the switch.

Which ports on the CX700 are currently connected to which hosts/HBAs?

I'm a bit confused by how you listed these ports.

You have two HBAs on each host, each one connected to one port on each of the two SPs (on the CX700).

So for example:

Host1_HBA0 connects to CX700spA_Port0
Host1_HBA1 connects to CX700spB_Port0
Host2_HBA0 connects to CX700spA_Port1
Host2_HBA1 connects to CX700spB_Port1

Leaving CX700spA_Port2 & 3 and CX700spB_Port2 & 3 open.

Yes, you can definitely manage this with no outage, although AIX servers have been "troublesome" in the past. As long as your CX700 has at least FLARE 19 on it and the hosts are running PowerPath v4.5 or greater (I have to double check the exact version, but if you are at the latest you don't need to worry), even AIX should be fine. You should try to minimize IO on each host as you are cutting over, but that is more of a CYA than a requirement. EMC will recommend you minimize IO if you are going to sit in front of the array and look at it for too long

Now, before you do this I'm going to give you one warning that you may or may not have considered. Right now you have full redundancy in your FC connections. By introducing a single switch (instead of two in dual redundant fabrics) you are introducing a single point of failure. While it is unlikely that an entire switch will fail, it is possible. Just something to consider and document to mitigate risk.

As for how to proceed, there are a couple of ways you could do it, but I'll tell you how I would plan it and you can go from there.

Step one: Get the four free sp ports (two per sp) connected to you FC switch. make sure everything looks good. This is also a good time to make sure you have four additional patch cables plugged into the switch and labeled ready to plug in spA0 & 1 and spB0 & 1. I'm also going to recommend that you set up zoning now to replace the physical connections you will be moving. If the Cisco is anything like the McData switches, it is easier to do AFTER you have a connection, but by doing it in advance you minimize the time you have no redundancy during the migration.

Step two: Connect your NEW host (the one that isn't SAN connected yet) to the switch and set up the zoning. Make sure the initiators show up on the CLARiiON and that you can allocate and see storage from the host. At this point you know all the infrastructure pieces work fine and you can proceed with migrating the existing two hosts.

Step three: On the CLARiiON force a trespass of all the LUNs for the existing two hosts to spB. PowerPath should handle this for you, but no need to take chances. Once everything is failed over, move your spA connection for Host 1 from spA0 to the switch and plug the patch cable from the switch to spA0. Set up the zone on the switch (if you didn't already) and verify that the initiator on the CX700 shows "logged in" again.

Step four: Assuming step three went fine, move your spA connection for Host 2 from spA1 to the switch and plug the patch cable from the switch to spA1. Set up the zone on the switch (if you didn't already) and verify that the initiator on the CX700 shows "logged in" again.

*** At this point you are halfway home.

Step five: Trespass all the LUNs for Hosts 1 & 2 from spB over to spA. At this point you are actually using the switch for all the storage traffic to the hosts. We just need to clean up the rest of the connections and you are home free.

Step six: Move your spB connection for Host 1 from spB0 to the switch and plug the patch cable from the switch to spB0. Set up the zone on the switch (if you didn't already) and verify that the initiator on the CX700 shows "logged in" again.

Step seven: Move your spB connection for Host 2 from spB1 to the switch and plug the patch cable from the switch to spB1. Set up the zone on the switch (if you didn't already) and verify that the initiator on the CX700 shows "logged in" again.

Step eight: At this point we just need to trespass all the LUNs back to where they started and make sure everything is running smoothly. Easy as pie and no outage!

I think that's it, but I'm open to suggestion and review from others before you take this as gospel. Any change of this magnitude should always be reviewed. And I put this plan together with a massive headache this afternoon. I'd love to hear confirmation from someone else before you try to use this plan.

You are correct that is exactly they are being utilized, I am sorry for the confusion I had to double check the port numbers and how they are currently being used. As for the migration do you think I am on the right course?