The thing I really like about the Fujitsu and Hitachi models is they are freaking cheap.

"Compared to the 2,000GB units available in the desktop
space, this appears almost pathetic."

Seagate Constellation ES is 2TB ...

Also, we working to design a 1.5 Petabyte storage system at work ... either drive form factor provides plenty of bandwidth when you use hundreds of drives. But the 3.5" drives offer much greater rack density ... nearly 100TB per 4U.

These are SAS drives but they test it on a motherboard with only SATA controller and I can't see any expansion cards in the hardware setup, what am I missing?

Me thinks they were trying to fulfill the dreams of some people using it in a desktop. They should have added Crysis load times since it is already in a desktop setup.

My goodness! Those drives need some cooling, or you could burn yourself on them!
Besides, high temperatures are not good for drives!

147 degrees! Wow And I thought my 1.5tb Seagates were hot at 105!

The SuperMicro motherboard has a built-in SAS controller. It has to, as SAS hard drives do not work with an SATA controller.

I doubt these tests used a SATA controller simply because the it would rather refuse to detect the drives.

No I/O charts? Charts for price per GB and per I/O would be nice.

I screwed up, the X8SAX does not have an onboard SAS controller- some other UP Xeon 35xx boards from SuperMicro do. Tom's had to use an add-in card somewhere as the ICH10R SATA controller in the X8SAX does not support SAS drives. I bet they used one of the SAS/600 cards they wrote a little article about earlier.

I would have loved to have the Intel x-25E 64GB drive to included as a reference to this benchmark. I know its not SAS, but the SAS controllers do support it so if the proper decision is based on performance it should not matter if its SATA os SAS.

Your test setup is missing the hardware you used to connect those drives to the rest of the test system as in a controller card or on board controller.

Good info.

Hmmm....For the IT environment it sounds very interesting. What about gamers and home users? What's in their future?

SSD's!

For those asking about their relativity to MLC & SLC based SSD.
Note, difference between different platforms & controller should be minimal when testing for pure disk performance.

Here are the typical numbers for Intel X25-E:
IOMeter 2003.05.10 Database - >4800 IOps avg. (drops to ~3200 @Q64)
IOMeter 2003.05.10 File Server - >3600 IOps avg.
IOMeter 2003.05.10 Web Server - >5800 IOps avg.
Source: http://www.tomshardware.com/review [...] 158-5.html

For Intel X25-M:
IOMeter 2003.05.10 Database - >3000 IOPs (Q1-Q16)then ~1000 IOps @Q64
IOMeter 2003.05.10 File Server - >1000 IOps avg.
Source: http://www.tomshardware.com/review [...] 158-5.html
PCMark Vantage HDD Suite - >28000 (applies to all Indilinx & Intel)
Source: http://www.anandtech.com/storage/s [...] =3631&p=24

For VelociRaptor:
PCMark Vantage HDD Suite - 6600
Source: http://www.anandtech.com/storage/s [...] =3631&p=24

Weight those numbers against numbers found in this review.

Basically the conclusion from the numbers I've gathered indicates the performance difference between 2.5" 15k vs. SLC-based SSD for database, file server and web server is day and night.
I find the continued offering of 15k RPM HDDs for enterprise usage is APPALLING even when $/GB has been factored in. The reason for slow uptake of SSD in those market is due to companies like EMC overcharging for SSD solutions.

And for those interested in desktop usage performance the difference between VelociRaptor and MLC-based SSD is selective in advantages. Look through individual HDD tests under PCMark Vantage to find out the difference in different usage areas.

I still have several raptors that have performed admirably but I think I will start to phase them out soon, I like the smaller form factor.

Well, with Corsair spec for its latest SSD being 100+ years of life, I just don't see the point in going via SAS vs SSD. Spinning platters at 15,000 rpm is going to wear bearings, something rotten, plus there is heat and power usage, whereas SSDs are low power and (almost) zero latency AND we don't have to purchase SAS controllers. They are also totally silent and there is no vibration. Plus SSDs are coming down in price. QED.

I think it would have been a good idea to start this article about 'what' SAS means or is. Its not mentioned anywhere in this article.

I love my two Fujitsu 300GB SAS 3Gb/s 15k rpm in RAID 0, but when I bought them I didn't realized that the connector is not like the SATA drivers. My motherboard is a MSI K9A2 Platinum and it includes 2 SAS ports and if you disable the Promise SAS controller in the BIOS you also disable the eSATA ports, which makes me wonder if they are eSATA or eSAS? Any way the thing is that the connection on this board is the same as SATA ports. I do have LSI MegaRAID SAS/SATA PCIe 4X Card and the cable that it came with it is a mini SAS connector with four SATA types connectors on the other side.
Does anyone know why almost all SAS Card come with this type of cables instead of SAS 29 pins cables?
Also I noticed that SAS drives have more data pins that SATA drives, actually twice data pins. And with the power pins added to the plug you can count 29 pins.
Here is my concern, does that mean that I can connect one of this cables with 29 pins to the SAS hard drive and 2 SATA like connectors on the other side of the cable to a motherboard, and I can boost the performance of the hard drive up to 6 Gb/s because of the dual connection? Does that mean that 6Gb/s SAS drives could perform up to 12Gb/s because of the same configuration with that type of cable?

what is SAS??