• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Raid 5 Performance?

J

Joony

Limp Gawd
Joined
Apr 30, 2002
Messages
495
59708250.jpg


This is a HD Tach benchmark for a server I just setup. It's a Dell PowerEdge 1800 with a Dell CERC 6 channel SATA Raid Controller (Rebranded Adaptec with 2 additional ports) along with 6 Western Digital WD3200SD (Raid Edition).

I configured the RAID5 array with a 64KB stripe size, Read Caching Enabled, Write Caching Disabled.

Does the performance look about right for this kind of setup?
 
That looks like about what I'd expect from that adaptec card. Old, out-of-date, dude, you're getting a Dell. :rolleyes: Those results are par for the course, but you might want to find a different course.

An Areca card will cost about $600 for 8 ports but it's so much faster that you might want to look into it.

 
So, I just ran the 'long test' on my array:
Promise SX4-M, 4x Samsung SP1614.
Keep in mind that I was still streaming MP3's off it at the same time.
hd_tach.PNG

that is my result.

The controller is in a 32-Bit PCI slot (not PCI-X) and I have a Celery 325 (2.533 GHz) with 1 GB of RAM (dell SC420)

so either your RAID card is really crappy (I think my promise stinks) or there is something fishy with the config.
 
Jeeze, I must have a really bad controller or something. Does it matter that I made two partitions? 40gb OS partition and a 1.1TBish data partition? It was also a fresh install of Server 2k3.

edit, it's also a 64bit PCI card in a 64bit slot.
 
Joony said:
Jeeze, I must have a really bad controller or something. Does it matter that I made two partitions? 40gb OS partition and a 1.1TBish data partition? It was also a fresh install of Server 2k3.

edit, it's also a 64bit PCI card in a 64bit slot.
Click to expand...

Nope it doesn't. I also have 2 partitions on the same array.
 
On Server 2K3 with 8x200GB Seagate PATA/100 drives on a Highpoint RocketRAID 404 I am able to get an average 66.1 MB/s for the short test. This card cost me less that $100 several years ago.

hdtach.png
 
That's typical for the Dell CERC family, which are crippled rebrands.
You're going to have to replace the card with something else.
 
I tested a spare drive (same as array drives) on another computer as a single disk. Blazing fast!

original.jpg


Yeah, I believe they are crippled. The official Adaptec model number is 2610SA. It happens that HP has its own version of the card as well. HP page on their "HP 6 Port SATA RAID Controller".

HP's current firmware version is: 4.2-1-7368
Dell's current firmware version is: 4.1.0.7417

Driver, HP - 4.2.1.7365
Dell - 4.1.0.7010

:mad:

I tried flashing the HP firmware into the dell but a few ID values didn't match up so the Adaptec flasher didn't let me update it. Maybe I can try to override with command line options, which exist apparently.
 
drizzt81 said:
So, I just ran the 'long test' on my array:
Promise SX4-M, 4x Samsung SP1614.
Keep in mind that I was still streaming MP3's off it at the same time.
Click to expand...
Off the same array you were testing? Then the results are completely invalid and should be discarded.
 
make sure you check the block size hd tach uses during the read test since different block sizes will give you different performance results on different stripe size arrays. for example on my raid 0 array with 4 wd2500ks i get 23.83 MiB/s on a sequential read of 4k blocks on a 16k stripped array, and 155.36 MiB/S reading 256k blocks on the same array with a stripe size of 16k. I know its not a raid 5, but these types of testing discrepancies might be giving you results that don't make sense, also the mhz, bit size, and raid scheme (hardware vs sw) will make a difference. 64 bit 66 mhz pic-x slots pump twice the data in half of the time compared to 32 bit 33 mhz pci-x. also if you card is a 32 bit card running in a 64 bit slot it will only run at 32. the considerations are a) controller specs, b) mobo bus implemenation, c) raid setup (hw vs sw). It's best to double check all the specs to make sure they all use the highest data rate your system can support. also check the hd tach and see if you can alter the block size to see if you can get different results. the real test would be to open 3-4 medium sized movies and play all the same time off the single array and see if you drop frames. that's a good test.

picture27mw.png
 
piako said:
the real test would be to open 3-4 medium sized movies and play all the same time off the single array and see if you drop frames. that's a good test.
Click to expand...
I have now stopped listening to anything you may have to say.

What system is this supposed to stress? Even the highest-resolution movies I have are >= 10 mbits. And it takes me 10% cpu to play one movie. Thus, if I open 10 movies, I'll have completely bogged down my CPU. And the disk system has to deliver 100 megabits per second - oh no how will I do that! That's 12.5 MB/s!

In short, this may be the worst test of hard disk speed I've ever seen. Even including Sandra. I congratulate you, sir; now kindly leave.

 
Performance seems abysmal, but it's hard to tell from the blurry jpegs :-p. Try som PNG lovvin'.
 
The graphs appear readable to me. What is it that you can't make out?
 
Someone just e-mailed me (they can't seem to post) that they have the same problem with the same card

I've run a number of a tests on a Dell 830 w/ a CERC 1.5 6-ch RAID using a different configs so I thought I'd share them with you. (Since I've been looking for info from others to no avail until I found your post.)

Basically, my findings showed that it didn't matter what configuration I was using, I would top out at around 52MB/s using HDTach. These configurations included dual 80GB Maxtors, quad maxtor 250's in RAID 0+1, RAID 5 and even RAID 0!

So the card is DEFINITELY the limiting factor, but your numbers DID look a bit low.

I hope that helps. If you find any resolution to this problem, please let me know. I'll do the same if your'e interested.
Click to expand...
 
Joony said:
Someone just e-mailed me (they can't seem to post) that they have the same problem with the same card
Click to expand...

That would be me... it took a couple of hours to get my registration to go through so I thought I'd help out in case it never happened.

Here are my graphs (both quick):

Single 80GB Maxtor:

single-80.GIF


vs

4x250GB Maxtor RAID 0:

4x250-RAID0.GIF


I also tried dual 80's RAID1, 2x250 RAID1, 4x250 RAID0+1, 4x250 RAID5, 4x250 RAID0. All of the RAID tests came out at around 52MB/s using HDTach. I was disappointed, but based on some of the problems I've seen other users report with some of the competitive products, I'm not sure I would want to switch at this point. (Plus I got it for free to do some testing since we sell them to some of our clients. I may think twice about that moving forward.)

All tests were performed using a 64-bit PCI slot in a Dell 830 on a server running 2003 Enterprise. (legit even!) ;-)

Just my $.02
 
You must log in or register to reply here.
Back
Top