 |
 | Home | Contact Us | FAQ | Search & Site Map | Link to Us |
 | Sign In | Join | Other 45 Sites in Network |
Windows Forum / Windows XP / General Topics 1 / May 2008Dual Disk SATA
Gurus, On a machine with two SATA drives, is there any performance gain by running applications off of the second SATA disk? I understand that SATA is serial in-line technology, that being said, if all the reads and writes have to come through the same controller, then perhaps running applications off of the second SATA disk might actually result in a DECREASE in total system performance?  SignatureSpin > Gurus, The hard drive ghurus are over in comp.sys.ibm.pc.hardware.storage . *TimDaniels* There are a number of hw items to be taken into consideration if seeking performance, not just the drives > Gurus, > [quoted text clipped - 4 lines] > applications off of the second SATA disk might actually result in a > DECREASE in total system performance? > Gurus, > [quoted text clipped - 7 lines] > -- > Spin You need to understand the way SATA drives are handled. Each drive is directly connected to a SATA port on the motherboard. They do not connect in a Master/Slave set up on the same cable. > Gurus, > [quoted text clipped - 5 lines] > applications off of the second SATA disk might actually result in a > DECREASE in total system performance? SATA interface speed is 1.5 or 3 GB/sec, harddrives deliver 50-70MB/sec. The drive will be the bottleneck, and two drives (in particular with the swap partition and personal files on the 2nd one, or even raid0/stripe) can enhance performance somehow. There is no sense in creating a separate partition for the swapfile on the first drive, since that would force excessive head movements across partitions. > Gurus, > [quoted text clipped - 4 lines] > running applications off of the second SATA disk might actually > result in a DECREASE in total system performance? Any inefficiencies or conflicts within the controller are insignificant and ridiculously small compared to disk access time. There's one technique, for example, that loads a program in disk-location order then sorts the various pieces out in RAM. MUCH faster than flopping all over the drive to load the pieces in logical order to begin with. If you want performance, raid/0 (Stripe) is the answer. Ramone > Gurus, > [quoted text clipped - 4 lines] > applications off of the second SATA disk might actually result in a > DECREASE in total system performance? > If you want performance, raid/0 (Stripe) is the answer. Yes it is. Just be aware that if either drive fails, all data is lost. If data is backed up like it should be, there's nothing to worry about. Ramone >> If you want performance, raid/0 (Stripe) is the answer. > > Yes it is. Just be aware that if either drive fails, all data is lost. > If you want performance, raid/0 (Stripe) is the answer. It produces very little additional performance, but greatly increases the risk to your data, since if any drive in a stripe is lost, all the data on the stripe is lost. I recommend against it. > > Gurus, > > [quoted text clipped - 4 lines] > > applications off of the second SATA disk might actually result in a > > DECREASE in total system performance? SignatureKen Blake, Microsoft MVP - Windows Desktop Experience Please Reply to the Newsgroup Why did my hard disk performance double? Sorry Ken but you are wrong on the performance issue. And like I said if data is backed up properly why worry? There is a much higher risk of losing data to a Windows failure than a hard drive failure. Yes a stripe setup does increase the risk but the trade off in performance increase is most definitely worth it. The only way I might not do it is in a business environment. Ramone >> If you want performance, raid/0 (Stripe) is the answer. > [quoted text clipped - 14 lines] >> > applications off of the second SATA disk might actually result in a >> > DECREASE in total system performance? > Why did my hard disk performance double? Sorry Ken but you are wrong on the > performance issue. That's your opinion. I disagree. I've tried RAID0 here and I saw no discernable performance improvement at all. I know many others who report the same thing. I don't know why your "hard disk performance" doubled, but I suspect that you simply didn't measure it carefully (if at all) and what you report is just wishful thinking. > And like I said if data is backed up properly why worry? I'm entirely with you regarding backup. That's a necessity whether or not you use RAID0. However, there is always a risk of backups not restoring properly, getting lost, etc. Having backups is great; relying on them to always be there working properly when you need them is not so great. And restoring from a backup takes time, and there can be a cost associated with that too. > There is a much higher risk of losing data to a Windows failure than a hard > drive failure. I don't agree, but it doesn't matter, because it's irrelevant. The point is simply that RAID0 increases the risk. And since it increases risk for little or no benefit, it's not worth taking the risk. > Yes a stripe setup does increase the risk but the trade off > in performance increase is most definitely worth it. Once again, I disagree. The performance increase is somewhere between tiny and non-existent. That's why I took it off my computer here. > The only way I might > not do it is in a business environment. [quoted text clipped - 19 lines] > >> > applications off of the second SATA disk might actually result in a > >> > DECREASE in total system performance? SignatureKen Blake, Microsoft MVP - Windows Desktop Experience Please Reply to the Newsgroup I don't know why performance did not double for you or anyone else. Maybe it's due to hardware issues. But I can tell you for a fact that my stripe setup on two different machines did factually double. I'm not trying to be difficult, I'm just giving my opinion based on my experience. Maybe you didnot use a hardware based raid? Ramone >> Why did my hard disk performance double? Sorry Ken but you are wrong on >> the [quoted text clipped - 56 lines] >> >> > applications off of the second SATA disk might actually result in a >> >> > DECREASE in total system performance? >I don't know why performance did not double for you or anyone else. Maybe >it's due to hardware issues. But I can tell you for a fact that my stripe >setup on two different machines did factually double. I'm not trying to be >difficult, I'm just giving my opinion based on my experience. Maybe you >didnot use a hardware based raid? how did you measure it - using what tool? > >I don't know why performance did not double for you or anyone else. Maybe > >it's due to hardware issues. But I can tell you for a fact that my stripe [quoted text clipped - 3 lines] > > how did you measure it - using what tool? Took the words out of my mouth! SignatureKen Blake, Microsoft MVP - Windows Desktop Experience Please Reply to the Newsgroup I used 3 different benchmarks, both before and after setting up the raid 0. HD Tune, HD Sentinel, and PCPitstop online test. On all 3 I went from 55-65 mb/sec transfer rate to 105-125 mb/sec transfer rate. Ramone >> >I don't know why performance did not double for you or anyone else. >> >Maybe [quoted text clipped - 8 lines] > > Took the words out of my mouth! >I used 3 different benchmarks, both before and after setting up the raid 0. >HD Tune, HD Sentinel, and PCPitstop online test. On all 3 I went from 55-65 >mb/sec transfer rate to 105-125 mb/sec transfer rate. > > Ramone So moral of the story is with SATA, hardware RAID 0 doubles performance (as you shown), software RAID 0 but doesn't offer very much except for a lot of potential risk (as Ken explained)! >> I used 3 different benchmarks, both before and after setting up the >> raid 0. HD Tune, HD Sentinel, and PCPitstop online test. On all 3 I [quoted text clipped - 4 lines] > So moral of the story is with SATA, hardware RAID 0 doubles performance > (as you shown), That is a very deceptive statement. Equating burst speed increases with actual performance increases is misleading, that is just not an accurate measure of performance! Most RAID-0 performance increases can usually be measured in the range of less that 5%. John >>> I used 3 different benchmarks, both before and after setting up the >>> raid 0. HD Tune, HD Sentinel, and PCPitstop online test. On all 3 I [quoted text clipped - 11 lines] > > John Actually the Burst rate increase would not be much better, it would be the sustained reads and writes that benefit from having the RAID 0. If you are doing say video editing then yes a gain would be seen. But little files that don't exceed the the onboard cache would maybe even drop slightly from overhead on the two drives. http://www.anandtech.com/storage/showdoc.aspx?i=2969&p=4 >>>> I used 3 different benchmarks, both before and after setting up the >>>> raid 0. HD Tune, HD Sentinel, and PCPitstop online test. On all 3 I [quoted text clipped - 18 lines] > drop slightly from overhead on the two drives. > http://www.anandtech.com/storage/showdoc.aspx?i=2969&p=4 Thanks for clearing that up, Bob. As you say, for video editing and the likes there may be real gains but for most other things the difference is negligible. The conclusion page of the same article pretty well sums it up: [Quote] If it is not obvious by now, RAID 0 will provide outstanding results in synthetic benchmarks but really does nothing in actual applications. We should probably clarify that statement in detail. Utilizing the best performing drives in RAID 0 is the setup to have if you are looking to publish top benchmark scores with results in PCMark05 improving by 25% as an example. That same setup will provide you with at best minimal performance improvements in most applications, or sometimes no difference at all. [end qoute] http://www.anandtech.com/storage/showdoc.aspx?i=2969&p=9 And from another article on the same site: [Qoute] If you haven't gotten the hint by now, we'll spell it out for you: there is no place, and no need for a RAID-0 array on a desktop computer. The real world performance increases are negligible at best and the reduction in reliability, thanks to a halving of the mean time between failure, makes RAID-0 far from worth it on the desktop. (...) Bottom line: RAID-0 arrays will win you just about any benchmark, but they'll deliver virtually nothing more than that for real world desktop performance. That's just the cold hard truth. [end quote] http://www.anandtech.com/printarticle.aspx?i=2101 Claims of performance increases in the 100% range are specious or dubious to say the least! John Those numbers are not for burst speed they are for average transfer speeds. The burst speeds are actually much higher but I pay no attention to those numbers. Ramone >>> I used 3 different benchmarks, both before and after setting up the raid >>> 0. HD Tune, HD Sentinel, and PCPitstop online test. On all 3 I went from [quoted text clipped - 11 lines] > > John Those numbers are for average transfer rates, not burst speed. I pay no attention to burst speed numbers. Ramone >>> I used 3 different benchmarks, both before and after setting up the raid >>> 0. HD Tune, HD Sentinel, and PCPitstop online test. On all 3 I went from [quoted text clipped - 11 lines] > > John It still makes little difference in the real world, these benchmark numbers do not translate into real and equal performance gains, at best (unless you work with really large files) the gains won't amount to anything much more than a couple of percentage points. For most users RAID-0 on the desktop is just not worth the added overhead and decreased reliability. John > Those numbers are for average transfer rates, not burst speed. I pay no > attention to burst speed numbers. [quoted text clipped - 16 lines] >> >> John There is always the "Placebo Effect" with many PC operations. First & foremost is the "Clean Install = Fast", which is mostly due to a platform which is not fully loaded. But there is a whole group of Windows users who follow a regular schedule of Format & Start over. RAID is just one setup that "most" desktop users do not need - the same for setting up & using Dynamic volumes. If the OP believes he gets a appreciable return using RAID-0 then that's fine, but he shouldn't globally recommend it to NG readers implying everyone will get an enormous performance gain. Anyone can post a favorite Tweak or Setup but should at the least take the time to mention any downside or risk with it. I have a system with 5 SATA drives and a couple of different controllers. I use both internal and external drives. Using DiskSpeed32, I measure an average throughput of ~82 Megabytes-per-Sec on all of them. This is more than enough speed. > It still makes little difference in the real world, these benchmark > numbers do not translate into real and equal performance gains, at best [quoted text clipped - 25 lines] >>> >>> John > >I used 3 different benchmarks, both before and after setting up the raid 0. > >HD Tune, HD Sentinel, and PCPitstop online test. On all 3 I went from 55-65 [quoted text clipped - 5 lines] > you shown), software RAID 0 but doesn't offer very much except for a lot of > potential risk (as Ken explained)! My personal experience was with hardware RAID. It provided no perceptible increase in performance. SignatureKen Blake, Microsoft MVP - Windows Desktop Experience Please Reply to the Newsgroup > Gurus, > [quoted text clipped - 4 lines] > applications off of the second SATA disk might actually result in a > DECREASE in total system performance? Can safely say that when I image my XP partition, and save that image to a partition on same hard drive it is relatively slow. Slower by about a 1/3 more time as opposed to when I save the image file to another hard drive. Both, are identical SATAs. SignatureDave > Gurus, > [quoted text clipped - 4 lines] > applications off of the second SATA disk might actually result in a > DECREASE in total system performance? Just incase some of the answers arent clear: In general adding a second SATA disk will never decrease your total system performance. Depending on how you use it you may well be able to increase total system performance. SignatureBrian Cryer www.cryer.co.uk/brian |
Reply to this Message |
 Sign In
 Join
 My Latest Posts My Monitored Threads My Blog My Photo Gallery My Profile My Homepage Start New Thread Enable EMail Alerts Rate this Thread
|
©2008 Advenet LLC Privacy Policy - Terms of Use
This website includes both content owned or controlled by Advenet as well as content owned or controlled by third parties.