sPECtre

10K V results posted

Recommended Posts

The new maxtor 10K are in the performance database, with stellar results, along with seagate cheetas 7K6, but in 73 G version, which seem to perform slower that their 146gb counterpart

Share this post


Link to post
Share on other sites

What's with the decrease in server performance for the Maxtor from the 10k IV? Are SCSI drives now starting to be optimized more for workstation/single user applications like ATA drives? Would seem to make sense since I doubt corporate America is pouring over drive benchmarks before make storage purchases. Maybe this next generation of 10k SCSI drives is going to be the response to the Raptor with hopes it will leak little into the enthusiast market the Raptor has had all to itself, while 15k will be the mainstream corporate money maker.

Share this post


Link to post
Share on other sites

Stellar results indeed.

What's with the decrease in server performance for the Maxtor from the 10k IV? Are SCSI drives now starting to be optimized more for workstation/single user applications like ATA drives?

That certainly seems to be a reasonable assumption. Of course, Maxtor has always optimized its SCSI products for the desktop more than the other SCSI manufacturers have.

Positional performance doesn't seem significantly impaired though. The drop is very small. I wonder if it might not be the firmware at all. It might be possible that the tiny 0.1ms increase in seek time could account for the equally small shift in IOMeter results.

I am now, very interested in the results for the next generation of 15K drives. This has also whet my appetite for the results of the 2.5" platter 10Ks. The next quarter should be a very interesting time at SR.

Share this post


Link to post
Share on other sites
Guest Eugene
That certainly seems to be a reasonable assumption.  Of course, Maxtor has always optimized its SCSI products for the desktop more than the other SCSI manufacturers have.

There may be another explanation... rather than overt optimization for single-user patterns, perhaps Maxtor has had a history of better TCQ implementation than competitors. Take a look at http://www.storagereview.com/test/test.html . With Seagate's recently discovered NCQ results, it's possible that NCQ itself is not necessarily superior in the incurred overhead department as many (myself included) were willing to assume but rather that Maxtor's implementation of it is. Maybe Maxtor's implementation of TCQ on the SCSI side has traditionally hampered its drives less than stuff from, say, Seagate.

Positional performance doesn't seem significantly impaired though.  The drop is very small.  I wonder if it might not be the firmware at all.  It might be possible that the tiny 0.1ms increase in seek time could account for the equally small shift in IOMeter results.

The slight increase in access time may do it. Going from 36 GB/platter to 73ish is a doubling, a huge increase compared to the relative baby steps that the ATA world has turned out in the last couple years. Combine that with the fact that our sample has the maximum head count of 8 and its not hard to imagine that certain things take just a wee bit longer.

Also, in the larger picture, keep in mind that our desktop measures are affected by capacity increases while the server tests are not. IOMeter can punish for tiny increases in seek time while in the real world a doubling of capacity could significantly reduce the average actuator travel distance.

I am now, very interested in the results for the next generation of 15K drives.  This has also whet my appetite for the results of the 2.5" platter 10Ks.  The next quarter should be a very interesting time at SR.

Some Savvio results have been in the database for a little while now. I haven't been able to get the drive to complete our desktop suite, however.

Share this post


Link to post
Share on other sites

The 10kIV samples I got all seem to be configured for a very high number of cache segments (226) while the older Atlas models always had just 21 (still more than most of the competition). I'd be interested in the settings for SR's 10kIV test samples.

Share this post


Link to post
Share on other sites

Gone are the days when you can test the biggest capcity drive of its family and have the same results for the rest. The quote and real-life access times and real life performance will vary depending on how many platters etc. The more heads, the more inertia to overcome and slightly slower read/write access times. In real life use, a higher capacity drive with less data (short stroke, greater/closer locality of reference) can counteract the slower quoted/tested access times.

Remember the trouble with WD drives with odd capacities (capacity not even .5 multiple of platter density).

Share this post


Link to post
Share on other sites
Any idea when we can expect to be able to purchase an Atlas 10K V or Cheetah 10K.7?  I thought they were supposed to be shipping in Q3.  We're nearing the end of Q3...

And when did Microsoft say Longhorn was going to be ready when they announced it? :) When HD mfr's say shipping in Q3, they actually mean the new drives may start trickling out to OEM's at the end of Q3 and won't be widely available to retail consumers until the end of Q4 :)

Share this post


Link to post
Share on other sites
Any idea when we can expect to be able to purchase an Atlas 10K V or Cheetah 10K.7?  I thought they were supposed to be shipping in Q3.  We're nearing the end of Q3...

And when did Microsoft say Longhorn was going to be ready when they announced it? :) When HD mfr's say shipping in Q3, they actually mean the new drives may start trickling out to OEM's at the end of Q3 and won't be widely available to retail consumers until the end of Q4 :)

From Hitachi's press release (February):

"The new 3.5-inch drives have already begun shipping and are in the process of being qualified at major OEMs worldwide. Hitachi expects volume shipments of the Ultrastar 10K300 to begin in the second quarter of this year."

From Fujitsu's press release (March):

"Qualification units for OEM customers will be available from April for the MAT series from June for the MAU series. Volume shipments will start shortly thereafter."

From Maxtor's press release (May):

"Maxtor Atlas 10K V and Atlas 15K II qualification units are scheduled for availability this quarter with full product availability scheduled for early in the third quarter of 2004."

So where are all these drives? History tells us no industry is worse at hitting their claimed release dates than hard drive manufacturers. I don't know why they even bother giving that information in their press releases any more. When you announce something will ship in volume within a couple months of your release, and you miss that date by months, you sure get the impression they were knowingly lying about the date, or are the worst time estimaters around.

Eugene: Savvio results are on the performance database? under what name?

It's under Seagate Savvio 10k.1. There are no workstation results posted for it.

Share this post


Link to post
Share on other sites

OEM customers, huh? That means I better talk to our purchasing department to see when these drives are really coming out. I think we qualify as a major OEM... :P

(I want my 10K V or a 10K.7...)

Share this post


Link to post
Share on other sites
Gone are the days when you can test the biggest capcity drive of its family and have the same results for the rest. The quote and real-life access times and real life performance will vary depending on how many platters etc. The more heads, the more inertia to overcome and slightly slower read/write access times. In real life use, a higher capacity drive with less data (short stroke, greater/closer locality of reference) can counteract the slower quoted/tested access times.

Remember the trouble with WD drives with odd capacities (capacity not even .5 multiple of platter density).

Is that to say they smaller the capacity, the quicker or "snapper" it will be? For instance, a drive that gets 300GBs from 4 (four) 75GB platters would be slower then that same drive with just one 75GB platter?

Share this post


Link to post
Share on other sites
Gone are the days when you can test the biggest capcity drive of its family and have the same results for the rest. The quote and real-life access times and real life performance will vary depending on how many platters etc. The more heads, the more inertia to overcome and slightly slower read/write access times. In real life use, a higher capacity drive with less data (short stroke, greater/closer locality of reference) can counteract the slower quoted/tested access times.

Remember the trouble with WD drives with odd capacities (capacity not even .5 multiple of platter density).

Is that to say they smaller the capacity, the quicker or "snapper" it will be? For instance, a drive that gets 300GBs from 4 (four) 75GB platters would be slower then that same drive with just one 75GB platter?

Yes, but only very, very slightly. Very likely, 0.1-0.2ms difference in seek time is all the variation you'd see in a given drive family.

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