Disk Arrays
cancel
Showing results for 
Search instead for 
Did you mean: 

Continuous Access and FCIP

Continuous Access and FCIP

I am getting ready to implement CA in our production environment in about two weeks. Due to budget limitations we only have a single T3 line between our sites. The primary site is in Hartford, CT and the remote site is in Austin, TX. The T3 line is a dedicated line for CA. We have (2) Cisco MDS 9222i switches with FCIP modules at each site (one per fabric).

Even though each 9222i switch has (4) FCIP ports, I am thinking of using only one per switch. Since the T3 line can only carry about 44Mbs, I don't think I would gain anything by using more than one FCIP interface port switch. In fact, I believe using more than one would saturate the T3 line.

Here is my question, when you configure FCIP on the MDS switches,you can specify the minimum and maximum bandwidth numbers for the tunnel. Since I got (4) links should I divide the T3 bandwidth (44Mbs) by 4 and essentially set the maximum number per switch at 11Mbs per FCIP port? Or should I just set them each to the maximim (44Mbs)?

Any suggestions or pointers are greatly appreciated. Also, we're doing asychronous replication and our average latency is about 40ms per trip.

Thanks,
Heather
1 REPLY
McCready
Valued Contributor

Re: Continuous Access and FCIP

I don't have the exact answer, but I would assume that your EVA will choose one path to take out from the controller via one of the two fabrics, so you will not be splitting your IO's between two fabrics. Thus, I would configure one FCIP port on each fabric to the maximum bandwith possible, as regardless, flow control of some type should keep the port from being overrun.

Those with better answers please respond...
check out evamgt.wetpaint.com and evamgt google group