File Server Performance

[djnes]

Okay, I'm asking for a ruling on which will offer better file server performance. The RAID controller is the same (Promise SX4000) in both situations, and there won't be any more than one or two file transfers going on at any given time. The server is the same in both situations, a dual 866 PIII, on a VP6, with 1 GB of memory. Here's the only differences:

A) 2 drives, both 120 GB WD SE drives in RAID1.
B) 4 Maxtor 20 GB drives in a RAID5 array.

I'm wondering if the RAID5 array will make up for the fact that the drives have less cache on each.

 

[DougLite]

The WD drives, hands down. You'd be lucky to get 50MB/sec sustained out of the Maxtor drives, which the WD drives should be quite capable of. The WD array also achieves a higher capacity.

 

[unhappy_mage]

What he said - 20gb drives have nothing to recommend them. Unless you meant 200 or 120, WD all the way.

 

[djnes]

You'd be lucky to get 50MB/sec sustained out of the Maxtor drives

So, being as it may that the Maxtor drives are so much slower, individually, even in a RAID5 array it's not enough to make up the performance differences?

 

[DougLite]

Yeppers. SR measured sustained transfer rates of about 15MB/sec for 10GB/platter drives. While those Maxtor units may be single 20GB platters, they still are not going to be significantly past 30MB/sec, and I doubt that striping them is going to more than double the throughput of a single drive.

 

[djnes]

Thanks....looks like I'll be doing a drive swap in my server soon.