PerfProb: 4x 15KSAS-> 6Gbs LexarCrucial->slowdown

Astara

Distinguished
Feb 3, 2009
21
0
18,510
I was looking to speed up my workstation -- I had 4 15K Fujitsu 147 (MBA314RC) configed w/RAID 0 under
my "Dell 7500's" builtin LSI-based SAS 6/iR" (BIOS RAID) controller). I got expected performance of about 4x a single drive. I short-strokeded them 10 .5 capacity to lower the seek time.

Win7 score: 6.1 (max read/write under windows ~ 500-550 MB/s).

A few years later, I found a fair deal with some good spec'ed Crucial (bought or owned by Lexar I guess),

The previous generation stated :(Read speeds up to 355MB/s);

(products:
cur: http://www.amazon.com/Lexar-Media-Crucial-2-5-Inch-CT128M4SSD2/dp/B004W2JKZI/ref=sr_1_6?ie=UTF8&qid=1311113837&sr=8-6
prev: http://www.amazon.com/Crucial-Technology-RealSSD-Solid-CTFDDAC128MAG-1G1/dp/B0039SM0AS/ref=pd_cp_pc_3
)

I figured even if I only got 200MB/s, that would be faster than my current SAS drives with lower seek!
(I was being very conservative in my expectations of 'improvement, I thought)...

Now, after much pain w/windows 7, in getting the new drives installed inplace of the old, (really didn't like me
substituting different drive types...)

I get a Win Perf score of 5.2 on the HD, and max speeds around , Write 426MB/s, read 515MB/s.

Not even close to 4x200, let alone 4x355, and certainly a drop below what I was getting with my
supposedly slower SAS drives.

I' noticed NO great speedup in boot times (the big promise of faster 'seek times')....

So what happened? Is the SSD hype just baloney (don't want to believe that, but was I fooled AGAIN?)...
or did I do something wrong in my upgrade/config?

The composite 'Dell Virtual Disk' size is the same as the total of the SSD's, so I'm sure they all being used,
and they ARE configed as RAID0, ...

Ideas?

Am tempted to go back to the SAS drives, cept' the SSD's do run cooler/quieter...but with a perf penalty...


 

tokencode

Distinguished
Dec 25, 2010
847
1
19,060
SAS has some additional enterprise class funtionality like command queing which can help performance. 4 15k SAS drives in RAID 0 is quite a nice setup, but I would arguue SSD is still faster/better valuef or the money for a workstation. 4 drives in RAID 0 will have 4 times the failure rate of a single drive. A single lrge fast SSD should in theory have a lower failure rate even though the drives you are using are enterprise class. It could have to do with th econtroller you're using for the SSDs. Have you tried stand alone?