difference between a director class SAN switch and a workgroup SAN switch ?

Yes, at least in the beginning.

With Virtual connect you'll need 25% of the host ports, so 110 -12 (14) = 98 (96) host ports.

98 (96) / 4 = 24 or 25 ports for hosts. Are your virtual connects 8Gb  ? In that case I'd think you'd only need 1 9148 switch.

Are the FA ports 2, 4 or 8Gb ? I'd put those ports to fixed xGb since you need guaranteed performance on those. The remainder of the switch performance is to be shared among the host ports. My gut feeling says you'll manage as long as your needed performance doesn't go up to much.

So 1 9148 per fabric and you're good to go.

Will the load the hosts put on the storage be the same as before ?

data "sheet" (movie): http://www.cisco.com/en/US/partner/prod/collateral/ps4159/ps6409/ps5987/ps10703/mds-9148-datasheet.html

on http://www.cisco.com/en/US/partner/products/ps10703/index.html it says it's a 48 port line rate switch, so non oversubscribed

48 x 8 = 384Gbps

I've read it somewhere, hang on, I'll look it up.

Take a look at Home > Products NULL&internalId=0b0140668002120d> > Hardware/Platforms NULL&internalId=0b0140668002120e> > Connectrix Directors and Switches NULL&internalId=0b014066800213c7> > Data/Spec Sheets and do a search on 9148.

It says the switch is capable of 8Gb non blocking so that would mean 48 non oversubscribed ports ! This is good news ! This is what Powerlin says. Let me verify this on the Cisco site.

Yes of course you'd need 2, 1 in each fabric.

Ah, I see posting per email isn't what I meant to show:

It should read Home > Products > Hardware/Platforms > Connectrix Directors and Switches > Data/Spec Sheets

Can you find this ?

I think so too. For redundance, I would by 2. The problem is I can't seem to find any rated data or actual data that would let me know how much data a workgroup class switch or a director class switch can handle to support my decision(which ever it might be).  (Like a LAN switch, it would say the number of packets it can handle per second) !

Thank you, I am able to find the page(s) watched/read the datasheet. I am not familiar with the term "non-blocking". Now, some what(thanks guys). Do I have to worry about if I have say 4 ports(from hosts) at full data rate hitting agains 1 Port (to Storage) that will bring this port down ?

non-blocking architecture means the switch is advertised to handle ..say 48 ports x 8G, it should be able to handle 384G of bandwidth without any delays.  Yes, multiple hosts running full speed will over-run your single port easily.

non blocking = line rate = non oversubscribed

You mentioned that now have an oversubscription of 1 to 7 (host to FA). So if the FA and hosts stay the same, nothing really changes. 10 or 15 to 1 is normal, I've even see 20 to 1, but those hosts didn't do much. I don't think you can brnig a port dwn, but if you have lots of heavy servers you might cause a 100% utilization of a oprt, causing it to appear slow. But really bring it down ? I doubt you can really do that.

Has anyone seen this happening ?