Slashdot Mirror
Flurry of Hard Drive Reviews
Sivar writes "After a long hiatus while setting up their new testbed, StorageReview.com has released a number of reviews of the latest hard drives, including Hitachi's Deskstar 7K500 which now occupies the top performance spot for desktop drives, the Seagate Barracuda 7200.9 which is the first shipping Serial ATA-II drive, the Seagate NL35 for backup servers and other "nearline" storage, and the Western Digital WD4000YR, which interestingly is actually based on their famous (and expensive) Raptor unit." Hitachi's SATA-II drive was also recently reviewed by BigBruin in case you missed it.
While high performance drives may be important, I feel that many drive manufacturers are forgetting that noise is a big issue. When you have 3 PC plus running at home high performance drives will just add to already high number of decibels that we have to suffer. Keeping my hearing is probably more important to saving a few milliseconds for each drive seek.
Karmady is the best medicine.
RAID improves throughput, but not latency. If you need low latency, you need high-RPM drives and no amount of RAID will help you.
The problem is that the density per disc is currently only about 125GB. So any drive over 125GB will have more than a single platter. The WD4000YR has 5 platters as do the Hitachi 400GB and 500GB. This extra mass is the problem which leads to higher temperatures, more noise and more power consumption.
160GB per platter is just on the horizon and with newer TUMR heads expect that to go up to 250GB per platter in a year. I wouldn't hold my breath for perpendicular recording. It still seems to be a couple years away.
Right now you can get 2GB flash chips. Put a few of these together and you have enough space for a decent install. The problem is that the performance is not quite there yet. You can expect raw read and right performance aroun 10MB/s on the fastest chips on the market. This is still alot less than the 50MB/s+ you find on desktop drives.
Samsung will do it before anyone else. WD, Seagate and Hitachi don't have any flash business so they are not going to push for it.
Slightly off topic, but Storage Review has always allowed the end user to compare any device to another by selection (eg 5400 rpm maxtor vs 10000 rpm WD) using discreet data fields (eg noise and heat). No other review site I know lets you do this, and its a very useful feature. Very often other review sites will scatter related devices across different non intersecting reviews, and I doubt they bother to break the data down to this level of detail.
"RAID improves throughput, but not latency."
Argh! This myth needs to end.
The only case in which RAID does not improve latency is that of a single tasking system.
The latency that's important for a multitasking system is the time an application has to wait for its data, not the time it takes the disk to process a single request. The benefits vary depending an access patterns, array geometries and RAID level.
Having more drives simply means there's a better chance that some requests can be handled in parallel. Your claim is akin to saying that people won't have to wait longer at the supermarket checkout when only one lane is open.
It's not a myth.
It's not an absolute, either, I'll grant you - but it is an excellent rule of thumb.
The only case in which RAID does not improve latency is that of a single tasking system.
This is not correct. RAID *might* improve your latency if its purpose is very specific, the setup can be carefully tuned for the access patterns and the physical placement of data on the disks is predictable, but in general it won't.
The latency that's important for a multitasking system is the time an application has to wait for its data, not the time it takes the disk to process a single request.
I'm confused. How isn't the time a disk takes to process requests directly related to how quickly the data can get to the application, in the general case ?
Having more drives simply means there's a better chance that some requests can be handled in parallel.
Certainly, but the chances of it happening are very low. A higher RPM drive will give immediate, predictable and consistent improvements in access times. A RAID array *might*, some of the time, if you're lucky and the planets are correctly aligned - but on average it will actually make latency worse.
Your claim is akin to saying that people won't have to wait longer at the supermarket checkout when only one lane is open.
Your analogy sucks. Not only is the scenario of people being served at checkouts talking about completely independent operations, but that independence also allows for performance hotspots (ie: longer queues in a particular aisle) to be avoided. Accesses to a RAID array exhibit neither of these characteristics.
Every Maxtor drive that I've bought in the last few years has died.
Disc drives are running hot these days. Heat is the enemy of a long service life.
Mount the drive so you can install a small fan in front, such as from an old power supply. Get a quiet fan with low vibration. Blow air across the top and bottom to cool the drive.
You should see a significant increase in reliability.
I see lots of benchmarking with and without NCQ enabled, and it appears that people are completely missing the point. If you have the option (and it works properly), you should never run a disk without NCQ! Benchmarking them together is a real disservice to all who don't understand the purpose of tagged command queueing.
NCQ allows the OS to know what has been committed to disk, which is very important from a reliability perspective. File systems do not function properly without this assurance, and can be seriously damaged on power failure.
To be fair, comparisons with NCQ should be made when write caching is turned off. Only in this case do you get the same level of reliability. Of course, ATA will be completely slaughtered, but it is a fair comparison. This abysmal performance led to the use of write caching; increasing performance at the expense of reliability. Now that it is possible to restore the reliability with NCQ, making a comparison without clarifying this point is not at all helpful.
The thing I would like to know is which disks actually implement NCQ properly, and which still lie to the system? Since drive manufacturers have been "cheating" for years on their IDE drives, has the situation truly been fixed? Spindle speed aside, it should now be possible to achieve the performance AND reliability that SCSI devices have offered for years. Unfortunately reviewers never seem to address this aspect.