SATA vs SCSI ...

Joel rees at
Mon Jun 27 07:16:57 GMT 2005

On Mon, 27 Jun 2005 10:13:04 +0400
"Artem Kuchin" <matrix at> wrote

> Marc G. Fournier <scrappy at> wrote:
> > looking at the specs between two cards, the SATA card(s) seem to rate
> > ~100-150MB/s on each channel (if I'm reading right), with both the
> > 3Ware and ICP cards having 4 individual channels ... looking at the
> > SCSI cards, they are rated at 320MB/s, but that is total for the SCSI
> > bus itself, right?
> > 
> > So, if I have three drives on  a SCSI bus, each 'maxing out evenly',
> > I'd be cap'd at about the same 100MB/s per drive, no?
> > 
> > In fact, looking at the SATA 2.x specs, each chanell there is rated at
> > 300MB/s, which, again, if I could 'max out evenly', could seriously
> > blow away the SCSI bus itself ...
> > 
> > *If* I'm reading this right ... ?
> For the last 6 month i really think that if you don't need something high-end
> scsi then you should go for SATA. There are test on sites such as
> Tom's hardware guide and

I'm always pretty cautious about how much trust I give those benchmarks
done by fanboy sites. Interesting stuff, but if it's mission critical
and benchmark is important, I'm going to design and run the benchmark
myself. And I'll cost that in when I give the estimate to the customer,
if the customer wants SATA.

(I think of a recent benchmark of Mac OS X/PPC and MySQL and nobody
mentioning sync issues in the benchmark article.)

> They show then on sequrncial read
> there is no difference between scsi and sata. Acatuallty, modern hdds use
> the same mechanics for sata and scsi versions of them. The brains (electronics)
> on the hdds are different of course. However, when it comes to random read/writes
> scsi wins because of command queueing. This was an issue until recently,
> Recently SATA with NCQ became widly available. Test show that some of those
> SATA disks with NCW ***WIN*** over scsi 320. The test envolve artificialy random
> read/write tests as well as real application benchmarking. I din't rememeber where
> excatly i saw the tests on those site, but you could search.

Well, I've been fighting on and off for about four weeks with an ATAPI
drive with SMART reporting bad sectors, and I'll offer this: when you
cost an ATAPI/SATA hard drive, make sure you factor in the cost of

Considering the business about Mac OS X and MySQL and sync, I'd factor
in UPS as well. Make sure you set it up to start taking the system down
long enough before the battery gives out that the drives have plenty of
time to clear their internal cache.

> So, my opinion, workstation never needs SCSI and every server MUST be 
> on mirror or RAID5 and there you should use SATA with NCQ drivers unless,
> your applicaton is really weird and needs something extremely speedy. Then, however,
> you could go for RAID 0+1 and get perfomance that SCSI will never get you.

Now that I have a bad taste in my mouth from SMART, I'm interpreting
that as "three SATA drives in RAID5 equals one SCSI drive". Contrary to
what you expect, I do not expect performance from that.

Joel Rees   <rees at>
digitcom, inc.   株式会社デジコム
Kobe, Japan   +81-78-672-8800
** <> **

More information about the freebsd-stable mailing list