Discussion Locked

RAID 5 Backup

0 helpful
2 replies
Closed to new repliesThis discussion is closed. It stays readable, but it takes no new replies.

Win 2000 Advanced Server, RAID 5, using BackupExec 8.6 tape deck on server. to back up 44Gb it is taking 16 Hours, now i know this is not right and have tried various options but still no joy. Has anybody experenced anything similiar and found theresolution. Whilst backing up at night, it seems that the CPU is max out at 100%. All I have come up with is possibly something to do with the RAID 5 system. But other RAID 5 servers backup ok.

Anybody......!!!!!!??

2 replies

by shanghai sam

Hi,

Run performance monitor and log lots of counters. Compare this with running perfmon against the server when the backup isn't running.

Hopefully you'll spot where the bottleneck is...

If its the backup software using all the CPU (and that is the bottleneck) then maybe there is a DMA setting you're missing, meaning your CPU needs to shuffle the data to the drive itself...???

Or is this particular server a file server with zillions of little files? It takes way longer to backup zillions of tiny files than a few big ones.

Is RAID 5 implemented in software rather than a controller? It could be very CPU intensive if the data is calculated from 4 disk reads rather than 5. This may be the algorithm used for zillions of smaller files...

Hope those thoughts help,
Cheers,
Tim.

by timthetoolman

Hi,

Run performance monitor and log lots of counters. Compare this with running perfmon against the server when the backup isn't running.

Hopefully you'll spot where the bottleneck is...

If its the backup software using all the CPU (and that is the bottleneck) then maybe there is a DMA setting you're missing, meaning your CPU needs to shuffle the data to the drive itself...???

Or is this particular server a file server with zillions of little files? It takes way longer to backup zillions of tiny files than a few big ones.

Is RAID 5 implemented in software rather than a controller? It could be very CPU intensive if the data is calculated from 4 disk reads rather than 5. This may be the algorithm used for zillions of smaller files...

Hope those thoughts help,
Cheers,
Tim.

ps. Sorry for the double post. TR times out if I take too long answering...

This discussion is closed to new replies.