|
Continued from Part I
I could go on and on with this but you get the general idea. If high availability is a concern, I generally use two fabrics, with two HBAs per server, one HBA to each fabric. In some cases, arbitrated loop is needed, especially if you have operating systems that do not support FC-SW protocol. FC-AL loops are generally MUCH cheaper to implement also. Arbitrated loops SHARE the bandwidth though, so if you are using a bunch of servers, performance may suffer, depending on the application load. You cannot share tape and disk devices on the same loop either, do to Lips on tape rewind operations. So you may need one loop for disks and one loop for tape access if you plan on doing SAN based backup.
EMC uses Connectrix switches in the SAN and they are McData based. McData switches do not provide for FC-AL connections like Brocade "Quick Loop" does. You CAN connect brocade switches and McData switches together now though, since the ratification of SW-2 protocol, which provides for inters, switch E-Ports from different vendors.
If you have more than 30 servers to connect than a high port count on the switch may make sense. This means using a director class switch, or using cascaded, looped, meshed or a "core-leaf" design for the fabric.
Building a SAN is a major investment. I would base my decision on Flexibility, Manageability, Availability, Performance, and Cost. A good method for determining who is offering the best solution for your environment is to either do a "bake-off" pilot test of all the vendors approaches, and see which works best, or temporarily hire an outside consultant do a "SAN ASSESSMENT" for your environment. It may be the best investment you make.
Chris
Editor's note: Do you agree with this expert's response? If you have more to share, post it in our Storage Networking discussion forum at http://searchstorage.discussions.techtarget.com/WebX?replyToMessage ((Content component not found.)) .8DmraOhxbkd^1@.ee83ce4!viewtype=convDate
|
