I got some HDD's SCSI U320's and they are NOT coming anywhere close to the specs via HD Tach. I called the manufacturer and found that they were DELL OEM drives (recent pulls). His thoughts were that they may have been tweaked for Dell. I know that Dell uses a lot of LSI Logic RAID controllers in their servers and I was using an Adaptec RAID controller. These are SCA drives so there are no pins on the HDD to jumper.
Has anyone heard of OEM's(like DELL), "tweaking of HDD's"? Are the controllers tweaked also? Is this simply a way of keeping you locked in to them for desired performance and a general HDD won't cut it on their systems?
The adapter I have for these HDD's has a jumper called synchronous. Any quick thoughts on if this can help?
Is HD Tach reliable when it comes to RAID evaluation? or is it best for single drives?
What's your adapter card? how about you post benchmark results?
If you're hard limited to 30-40MB/sec, it's an interface issue... you're running Single Ended when you need to be running Low Voltage Differential.
Cable: is it a flat cable? (Y) [LVD cables use twisted pair]
Terminator: is it LVD compatible? (N)
Drive: Is it configured to run in SE only mode? (Y)
Host adapter: Is it set to run in SE mode? (Y)
If any of the answers matches what I've put, you're running in SE mode.
The cable is twisted rated for 320MB/S
The card is an Adaptec 2200s/64 PCI-X
The terminator makes no difference I have used diff ones with the same results
I am limited to about 30-40 in Raid Mode about 60 in burst
There is nothing on the controller to change or the driver that I can see so it may be fixed by the OEM--BUT wouldn't this be limiting the throughput to fix it with se?
Asynchronous mode = OLD 5MB/sec SCSI. Don't turn it on, leave the bus synchronous.
Even though the drives are SCA, they probably have jumpers on the PCB of the drive, located opposite from the SCA connector. Check to be sure that the "SE ONLY" jumper is not active.
This is on an ASUS MB P5WDG2-ws-pro an Adaptec Dual Channel 2200s/64 in a PCI-X slot. The drives are fijitsu mas3184nc. There are no jumpers on the drives.
If Async is the default then I could jumper the SCA PCB's to SYNC and see if that works. *I did it but no difference*
I am getting about the same with 2 drives stripped as with one alone.
Single drive = Burst stripped is 135MB/sec sustained about 59 MB/sec
2 stripped = burst 138MB/sec and sustained 69MB sec
4 stripped = burst 139MB/Sec and sustained 69MB/sec
So no benefit from 4 drives stripped
RAID 10 is when it goes into the tankI'll do a bench on that later
Wondering if I put 2 drives on each channel if that would help. Dont want to do that because then I would have to put another scsi card in the system to support the U160 drives I currently have on channel 1 of the 2200S
Trying to find a DELL PERC4e (LSI 320-2E)card to see if anything changes. The Adaptec storage manager does not help
Yep Wusy I tried that. They said they had none to send. I had asked them if it would make a diff because I have 2 different revisions of the same drive (3 of each) their answer was no.
Don't know if they were telling the truth because these were oem'ed for Dell. I live in Austin, TX so a lot of dell parts are avail cheap.
will the cache throw off the readings? It has to go through the 64MB cache first and the overflow then gets measured. During this time lag the burst and avg speeds woube downgraded. Am I thinking right on this? Is that why basically a 2HDD raid stripe is so close to a 4HDD raid stripe?
You are about to answer a thread that has been inactive for more than 6 months. If you still wish to proceed, please ensure that your posting is original and does not duplicate or overlap any prior responses to this thread.