|
Yup, you need the multi-path drivers.
When configuring your SAN everything should be redundant. The FAStT700 has redundant controllers and the drives failover to the surviving controller should a controller fail. This is what I call "disk side failover". In order to provide complete redundancy, you should also have "host side failover." The ability to automatically fail over to a surviving path, should you happen to lose an HBA, switch, cable or controller, requires two separate fabrics, each hooked up to a different HBA in the host and connected to separate controllers on the storage array.
If an HBA fails, the path failover software automatically transfers all I/O to the surviving path. If a switch fails, or a cable is pulled, the failover software detects the failure and again, moves I/O to the surviving path. If a controller fails, the storage array connects the LUN to the surviving controller.
You need to be aware of issues with operating systems like Windows when using two paths to the same disk without a filter driver (which is what the path failover software is). If you hook up both controllers to a single fabric, to a single HBA in the server, disk administrator will list the drive twice. This can be a real pain in the you know what since only one path is the "real" path during normal activity. The pathing software will "filter out" both paths and disk admin will only see a single "logical" path to the same LUN. Most current pathing software also allows you to load balance your I/O across both PHYSICAL paths (using the single logical path). This means that if you are using 2-gigabit Fibre Channel you can get 400MB per second to a single LUN (providing the storage array supports active/active paths).
There is a good paper from IBM called the "IBM SAN Survival Guide" that discusses best practices on implementing an IBM SAN.
Chris
Editor's note: Do you agree with this expert's response? If you have more to share, post it in one of our discussion forums.
|
