IDE, SATA, RAID0, RAID4, and RAID5

[me at mattgillen.net (Matthew Gillen)]

markw at mohawksoft.com wrote:
>> On 2/6/07, markw at mohawksoft.com <markw at mohawksoft.com> wrote:
>>> I am lucky enough to have a three mechanically identical drives, one
>>> PATA
>>> (EIDE) and two with SATA and took the time to make a few tests:
>> Cool test!  Where's the read performance benchmarks??? ;-P
> 
> I didn't, in fact, do read performance. I didn't think it necessary for a
> couple reasons:
> 
> (1) There was nothing on the drives to read.
> (2) Read is almost always faster than write and write is usually the least
> cached data transfer, so, to me at least, write would seem to be the most
> indicative of real performance.

I would run Bonnie++ if you really want to do some performance benchmarks (
http://www.coker.com.au/bonnie++/ ).  It measures more than throughput (such
as CPU usage) and does different types of reads and writes (ie sequential,
random) so that you can see how it would perform for some real-life
applications.

The dd technique may get you strange results for a few reasons including the
lack of going through the filesystem code (your 'of' was the raw device file).

Matt


-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.