One of our clients with a rather large SBS installation had a drive failure.
Between the phone call about the beeping in the server closet and our monitoring we were aware of it quite quickly.
This particular SBS box has a 10 drive hot swap setup in an Intel SC5400LX series chassis. The RAID controller behind everything is an Intel SRCSAS18E PCI-E 8x.
The data RAID 5 array had a dedicated hot spare. However, the drive failure was on the OS RAID 1 mirror.
So, we dropped out the dedicated hot spare and allocated it to become a global hot spare. Once this happened, it automatically dropped into the degraded RAID 1 array and began to rebuild.
Now, the last time we had this happen, we were dealing with a 750GB Seagate ES.2 on a RAID 1 array (previous blog post). In that particular instance the rebuild took all of 4-5 hours.
So, since we were dealing with a 250GB drive, the assumption was that the rebuild would take no more than a couple of hours.
This assumption was based on the fact that the two controllers were similar in their configuration, with the SRCSAS18E PCI-E 8x connector providing more bandwidth capability than the SRCSASRB's PCI-E 4x.
This was not the case. Apparently the on board processor may have a lot to do with it as the rebuild took 24 hours.
Given that experience, the 750GB drive that failed on the other system would have taken 3 days to rebuild!
We are fortunate that this client is due for a refresh. The above box will receive a newer S5000PSLSATAR, dual E5440 Xeon Quad Cores, 16GB ECC RAM, and an SRCSASRB RAID controller. All of the drives will be renewed and a slightly different RAID array setup will be put in place. We are looking for data throughput, so the large array will be a RAID 0+1 setup.
It will be then provisioned as a backup data mirror for the main company shares as well as a Hyper-V box for some desktop OS VMs.
Microsoft Small Business Specialists
*All Mac on SBS posts are posted on our in-house iMac via the Safari Web browser.