snomunke

Raptor 150 doubter

Recommended Posts

I was going through my desk drawers found some nice ripe flame bait, so I thought I'd toss it out before it spoiled.

Now, I will preface my comments with this: I DO NOT own, have NOT physically seen/touched/smelled, seen in action, etc. the new Raptor 150 drive (or its younger brother, the new vaporware 74GB version). I am basing all my observations on what I have read here in SR and elsewhere. I do have the older Raptor 74GB (00FLC0), so I am using that as a stand-in for the Raptor 150 (wait...hold the flamethrowers a few more minutes...I will explain.).

OK, so this is it... I was currently testing the Adaptec 4805SAS controller with a Hitachi 36GB 15K SAS HDD (OS was loaded on this drive). I was having some problems with the controller and was getting very frustrated with troubleshooting it so I pulled it out and reload the OS on the Raptor 74GB. And WOW! did I notice a difference. Doing day-to-day computer stuff (i.e. booting, loading programs, surfing, music, etc.) the SAS drive was LIGHTYEARS faster than the Raptor 74GB.

Yet, the SAS drive is in the same Ultrastar 15K147 family tested by SR (SR tested the 147GB Ultra320) and shows up in the SR database at a significantly lower performance level than the Raptor 74GB in single-user desktop usage.

So, I am wondering...was the SAS drive so much faster because it had the advantage of the SAS controller's 128MB cache? Or is it just a faster (higher performance) HDD? Does the SAS interface possibly play a role? Bottom line is this SAS drive was MUCH faster in a single-user environment than the Raptor 74GB...I seriously wonder if the Raptor 150 could top it....

OK...I've donned my asbestos jacket...light the flamethrowers...

Share this post


Link to post
Share on other sites

OK, I'll bite. I just got a 150 Raptor and am wicked happy with the speed boost. Also the level of quiet, except when it seeks, but I never did mind that noise, kind of like it :) So, simply send me a controller with 128MB cache and I'll hook it up and send you the results :)

Dave

Share this post


Link to post
Share on other sites

I am have no doubt the Raptor 150 is fast. But when I see how much faster my 15K SAS drive is in a single-user environment over the 74GB Raptor, I can't imagine the Raptor 150 being faster than SAS. I am hoping someone here has experience with both the Raptor 150 and a SAS HDD (or at least a fast U320 HDD) and can tell me the Raptor 150 really is king...

Controller will be in the mail as soon as $1000 is Paypal'ed to me :-)

Share this post


Link to post
Share on other sites
I am have no doubt the Raptor 150 is fast. But when I see how much faster my 15K SAS drive is in a single-user environment over the 74GB Raptor, I can't imagine the Raptor 150 being faster than SAS. I am hoping someone here has experience with both the Raptor 150 and a SAS HDD (or at least a fast U320 HDD) and can tell me the Raptor 150 really is king...

Controller will be in the mail as soon as $1000 is Paypal'ed to me :-)

I need to be in complete denial about anything being faster than my new hard drive for at least 2 weeks. And... I'll get right on that Paypal thing, as soon as I quick e-mail Jennifer Anniston and get a date with her )

Dave

Share this post


Link to post
Share on other sites

Did you try both drives on the same controller? As SAS controllers should work with SATA drives, this should work. Could be your SAS controller has better drivers. Or the SATA drive could be badly fragmented and the SAS drive not.

Share this post


Link to post
Share on other sites

Well SR hasn't tested any SAS drives yet. I believe SAS has a lower command overhead than SCSI and perhaps that could result in much better performance. That report that HP did on their SAS versus SCSI showed a significant improvement in their RAID system's performance. I have no idea how that would compare to a single drive doing single user workloads though...

So who knows, you could be right. There is no SR data to dispute your claim that the SAS drive is faster than the Raptor.

Share this post


Link to post
Share on other sites
Most of the performance incearse is 15K vs 10K. Try running 8, 15Ks in RAID 0, that will open your eyes. :)

To what? How quickly your array can evaporate before your eyes?? :lol::rolleyes:

Seriously, RAID 0 is all but useless except for a very few, highly specific purposes. If you've got 8 drives, I'm willing to bet that you could configure them to perform much better than just putting them all in RAID 0.

Edited by Trinary

Share this post


Link to post
Share on other sites

Most of the performance incearse is 15K vs 10K. Try running 8, 15Ks in RAID 0, that will open your eyes. :)

To what? How quickly your array can evaporate before your eyes?? :lol::rolleyes:

Seriously, RAID 0 is all but useless except for a very few, highly specific purposes. If you've got 8 drives, I'm willing to bet that you could configure them to perform much better than just putting them all in RAID 0.

LOL. I have never, ever had a raid 0 array break on me. Never. Im not running it any more but I was for years. Plus it was only my OS, not my critial data that must stupid users decide to put on raid 0 arrays. Plus, I had more then 8 drives. I had seprate arrays for my data and backups. For what I did/do, raid 0 did wonders because I need high STR. Everyone has always told me exactly what you did yet none of them have had such an array or had the need for one.

Share this post


Link to post
Share on other sites

Most of the performance incearse is 15K vs 10K. Try running 8, 15Ks in RAID 0, that will open your eyes. :)

To what? How quickly your array can evaporate before your eyes?? :lol::rolleyes:

More like how quickly your wallet can evaporate before your eyes! :D

amdoc2, you're right, of course. RAID 0 is practically useless compared to separating concurrently accessed datasets across the same two drives, but if you've already separated your concurrently accessed datasets, then adding drives to put each dataset on its own RAID 0 array is a valid way to boost performance (assuming cost, drive bays, heat and noise aren't major factors).

I'll second HMTK's suggestion to try the Raptor on the SAS controller. I've heard big controller caches can make for a big performance boost. Most SATA users wouldn't buy a 128 MB cache SAS controller just for that boost, but since snomunke has already got one, I'd be interested to see if it boosts the smoothness on the Raptor to similar or greater levels.

Share this post


Link to post
Share on other sites

I will bite!

Here is my box of Raptor 150's.

http://www.realworldbenchmarks.com/forums/showthread.php?t=3

I also have two Seagate 15K SAS drives and 2 500 GB Nearline drives. I plan on doing several tests on a couple of different SAS and SATA RAID conrtollers. The controllers will be here next week so I am sitting blind to the SAS stuff right now.

As far as the 150GB Raptors go, two of them will max out an NVIDIA RAID controller built into dual Opteron boards. I am in the middle of testing 5 of them right now for a review and I am currently 3 deep into it. Three and four drives do not show an improvment over two drives.

Anyhow, I will report back once I am finished up with my review and everything is posted.

Share this post


Link to post
Share on other sites

It's in your head. With no direct comparison, and no hard numbers, just going by it feels faster than a slower model is about as poor a comprison as you can come up with. If there was any real difference, the 128MB caching controller is almost certainly the reason.

SR did test a SAS version of the 15k.4 I believe, and there was no real difference between it and the P-SCSI version. Or maybe I'm remembering something that didn't happen, since I can't find the article.

LOL. I have never, ever had a raid 0 array break on me. Never. Im not running it any more but I was for years.

And in a related story, 100% of people regardless of age who died in a car accident last year, had never had a fatal accident before. Past successes are not an indication of future successes.

Share this post


Link to post
Share on other sites
It's in your head. With no direct comparison, and no hard numbers, just going by it feels faster than a slower model is about as poor a comprison as you can come up with. If there was any real difference, the 128MB caching controller is almost certainly the reason.

SR did test a SAS version of the 15k.4 I believe, and there was no real difference between it and the P-SCSI version. Or maybe I'm remembering something that didn't happen, since I can't find the article.

LOL. I have never, ever had a raid 0 array break on me. Never. Im not running it any more but I was for years.

And in a related story, 100% of people regardless of age who died in a car accident last year, had never had a fatal accident before. Past successes are not an indication of future successes.

Correct, but like I said, it is not a "fatal accident" when it happens. Its not a big deal to reinstall windows. Or I guess its a huge task for some who dont know how.

Share this post


Link to post
Share on other sites
It's in your head. With no direct comparison, and no hard numbers, just going by it feels faster than a slower model is about as poor a comprison as you can come up with. If there was any real difference, the 128MB caching controller is almost certainly the reason.

SR did test a SAS version of the 15k.4 I believe, and there was no real difference between it and the P-SCSI version. Or maybe I'm remembering something that didn't happen, since I can't find the article.

LOL. I have never, ever had a raid 0 array break on me. Never. Im not running it any more but I was for years.

And in a related story, 100% of people regardless of age who died in a car accident last year, had never had a fatal accident before. Past successes are not an indication of future successes.

"...just going by it feels faster than a slower model is about as poor a comporison (sic) as you can come up with." hmmmm...

well, since I am on a computer 18-20 hours a day, I think I am qualified to use "feeling" as a benchmark...

Share this post


Link to post
Share on other sites
well, since I am on a computer 18-20 hours a day, I think I am qualified to use "feeling" as a benchmark...

ROTFL we have a saying here - even a donkey does not hit its nose on the same rock twice !

Yup, you asked for the flamethrower..... CATCH ! :D

Share this post


Link to post
Share on other sites
since I am on a computer 18-20 hours a day

If I were to use a pc for hat long every day the only thing I'd feel would be a throbbing headache...

Share this post


Link to post
Share on other sites
since I am on a computer 18-20 hours a day

If I were to use a pc for hat long every day the only thing I'd feel would be a throbbing headache...

got one of those too! LOL

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now