UNSOLVED
'Background Initialization' with PERC4 and PV220S
Splitting off from another thread, I have a question about what is happening when a PERC4 and PV220S array perform background initialization (BI). How long should it take, and what sort of penalty does it impose on performance?. Does anyone have any real-world experience with this?
After several hours with support trying to figure out why our SCSI arrays are so slow, and after our larger array (7x2x300GB, RAID5) had spent more than 90 hours to reach 81% complete on BI, support suggested that we abort the BI, tear down the virtual disk and recreate it, and perform the initialization through the BIOS utility, which takes "just a few seconds".
Right, we did that and indeed the utility reported 100% complete in about 20 seconds. Amazing! The we went into Windows to create the stripe set... and found that the BI process has begun all over again. It will be next Tuesday before it completes, since it's going at something less than 1% per hour.
Can this possibly be normal and expected?
Meanwhile the array that had already completed its BI is still pig slow, and very erratic in performance, showing periods of little or no activity, intermixed with periods of sub-par activity. On average or at peak, it performs worse than the system SATA drives.
Could this be a bad PERC card? Two separate bad or misconfigured arrays? There are no events, no alerts, just terrible performance.
After several hours with support trying to figure out why our SCSI arrays are so slow, and after our larger array (7x2x300GB, RAID5) had spent more than 90 hours to reach 81% complete on BI, support suggested that we abort the BI, tear down the virtual disk and recreate it, and perform the initialization through the BIOS utility, which takes "just a few seconds".
Right, we did that and indeed the utility reported 100% complete in about 20 seconds. Amazing! The we went into Windows to create the stripe set... and found that the BI process has begun all over again. It will be next Tuesday before it completes, since it's going at something less than 1% per hour.
Can this possibly be normal and expected?
Meanwhile the array that had already completed its BI is still pig slow, and very erratic in performance, showing periods of little or no activity, intermixed with periods of sub-par activity. On average or at peak, it performs worse than the system SATA drives.
Could this be a bad PERC card? Two separate bad or misconfigured arrays? There are no events, no alerts, just terrible performance.
