SATA RAID throughput question

A question to the SATA RAID Experts:

We measured the throughput for an INTEL SCRS28X SATA RAID controller with a few SATA II drives and determined, that we realize a data throughput (all cache modes disabled -> write through) from about 0.2 upt ro 0.8 MBit/s. The same drives connected to the onboad RAID controller results up to max 3 MB/s (also cache disabled). This happens also in modes without any RAID. Which cache mode enabled, the drives have a expected throughput from about 60 MBit/s on both controllers.

Does somebody have an explanation for me, why the throughput is such poor if caching is disabled?

We measured with benchmarking tools, but also if you just copy a file around, you "feel" the poor performance.
LVL 35
BembiCEOAsked:
Who is Participating?
I wear a lot of hats...

"The solutions and answers provided on Experts Exchange have been extremely helpful to me over the last few years. I wear a lot of hats - Developer, Database Administrator, Help Desk, etc., so I know a lot of things but not a lot about one thing. Experts Exchange gives me answers from people who do know a lot about one thing, in a easy to use platform." -Todd S.

The_IT_GarageCommented:
Read the "Disk Cache" portion of this page: http://tinyurl.com/lamof

It explains what you're seeing as well as being a good information of what caching of all types is.
0
jamietonerCommented:
I cant find any info on an intel scrs28x, but if its a standard 32-bit pci card that would explain the slow throughput (133MB/s) its being bottlenecked by the slower pci bus (sata II hdds have a max theoretical throuput of 300MB/s), the onboard controller isnt going to be bottlenecked like the pci card so you will get faster performance. The fix for this would be to get a pci express raid controller (4X pci-e 1000MB/s) or a PCI-X (532 MB/s) controller.
0
Renato Montenegro RusticiIT SpecialistCommented:
Please, mount only one disk and test it with HDTach (freeware):

http://www.simplisoftware.com/Public/index.php?request=HdTach

Run the long test. You should achieve something beyond 48 MBytes/s on average read. Maybe more, maybe less.

After that, build the RAID. and try again. You wont double the drive's transfer rate but must get almost twice as fast - 80 to 95 MBytes/s - in a RAID 0 (for example) setup.

Lower speeds or results that vary in each pass of the tests, may be due to a defunct hard drive or controller. In this case, test each item at a time.
0

Experts Exchange Solution brought to you by

Your issues matter to us.

Facing a tech roadblock? Get the help and guidance you need from experienced professionals who care. Ask your question anytime, anywhere, with no hassle.

Start your 7-day free trial
Ultimate Tool Kit for Technology Solution Provider

Broken down into practical pointers and step-by-step instructions, the IT Service Excellence Tool Kit delivers expert advice for technology solution providers. Get your free copy now.

Renato Montenegro RusticiIT SpecialistCommented:
One more information. Cache should improve the performance. "Write Back" is the best configuration, but if you dont have a UPS device, it can be too risky.
0
BembiCEOAuthor Commented:
Controller is
http://www.intel.com/design/servers/RAID/srcs28x/index.htm
8 channel SATA II RAID controler - PCI-X 133 Mhz

Onboad is Intel E7230 Chipset with LSI Embeded with Intel 3,0 GHz P4 (Prescott)

What I'm wondering about are two things:
Both controllers are affected, less than 3 MByte/s is not really what I expect (1 single HD on internal or external controller).  
Both controllers results in expected rates, if cache is enabled.

Everything I have read about different caching technologies is talking about a vaiance between 30 - 50%. The hardrive delivers data about 60 MBytes/s, that means, the throughput should be round about 30 MByte/s in worst case.

The benchmark I have used is PassMark Performance Test, which uses a 1Gig file, so the 128 MB Cache of the controller should not make such a big difference in performance.  

I used HDTach on my system (E7230 Chipset, 3,2GHz Prescott, LSI Embeded Onboad Raid 1 with 2xMaxtor DMax10 and a Adaptec U320 SCSI Controller with 3 Maxtor Atlas 10K), all cache disabled

On the LSI Controller, I get av. 59,2 MByte/s (165 Burst) (Raid 1)
On the SCSI Controller, I get 72,3 MByte/s (207 Burst) (Raid 5)

Which looks o.k.

As the problem machine is similar than my machine, it should show similar results?


0
Renato Montenegro RusticiIT SpecialistCommented:
Yes, it should.
0
scrathcyboyCommented:
The cahcing is a motherboard/CPU caching ,and it makes a 10x to 50x improvement in performance, the specs you got are software caching, hardware caching on the CPU and MB make a HUGE difference in performance.  ANy 3000 CPU today, if you disable the cache on the MB, it will run like a 350MHz CPU.  Caching at hardware level is most important for file reads, copies and transfers across the data bus.  If you dont have this enabled, simple file copying, like in windows install, takes 20x to 50x longer than cache.

Conclusion, hardware caching at the board level is what gives you the performance you expect in file copies across the DATA bus of the motherboard, without it, the speed is AT LEAST 10X slower, more likely 20X slower.
0
Renato Montenegro RusticiIT SpecialistCommented:
Sorry scrathcyboy. The cache in question is the disk cache. Not the CPU cache. But, even with disk cache disabled, the throughput should be in the range shown in his last post.
0
It's more than this solution.Get answers and train to solve all your tech problems - anytime, anywhere.Try it for free Edge Out The Competitionfor your dream job with proven skills and certifications.Get started today Stand Outas the employee with proven skills.Start learning today for free Move Your Career Forwardwith certification training in the latest technologies.Start your trial today
Storage

From novice to tech pro — start learning today.

Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.