Enlightenment

Member
  • Content Count

    40
  • Joined

  • Last visited

Community Reputation

0 Neutral

About Enlightenment

  • Rank
    Member

Contact Methods

  • Website URL
    http://

Profile Information

  • Location
    The Netherlands
  • Interests
    Storage, RAID, NAS, FreeBSD
  1. Enlightenment

    Raid 0 performance

    Thanks for pointing that out, i didn't know Vista finally corrected this easily avoidable misalignment. I'm puzzled by your 'more often aligned' though. If it aligns on exactly 1024KiB offset it should always be properly aligned, with any stripesize up to 1MiB. Can you clarify? Anyway, that still means anything below Vista by default has the worst possible misalignment on any RAID - software or hardware. Windows also does not allow 'Dangerously Dedicated' mode, in which no partitions are used at all. Also, many people don't know about this misalignment problem and how to correct it. They end up having a volume that may be fast on sequential transfers (due to buffering and non-contiguous access, the misalignment doesn't do much harm). But their IOps are crippled, especially RAID5 random write ends up far worse than necessary. And IOps are still the most important for both desktop and server systems. What exactly don't you agree on? A 1MiB offset should provide perfect alignment for stripesizes up to 1MiB, no matter what RAID level you use. RAID1 is exempt because it has no stripes and thus no misalignment. As for stripesize, as you said ideally you want one virtual I/O to be processed by one physical hard drive. A too low stripesize (and a misalignment) could mean multiple disks are working on one virtual I/O request. Some believed this to increase sequential throughput, but its far better to have a large stripesize so all random reads and writes can be handled by one physical hard drive, and the other disks in the RAID can be occupied for other I/O requests. This only works if there are multiple I/O requests, so some applications using "blocking I/O" read requests won't see the full potential of the RAID volume. This is often noted in HDTune, which only used a queue-depth of 1 to access the device, bypassing all the optimizations the filesystem does. By using ATTO to benchmark your RAID you see what you would truely be getting when operating on the filesystem level. This is because the filesystem uses buffering, write-back and read-ahead mechanisms, which send multiple I/O requests at once. I expect you to know this, but i hope its an interesting read also for others. I understand what you mean, but i think this is wrong. Why should the full stripe block (or chunksize as you call it) be aligned? Normally only the stripe cells should be aligned, as in: | DISK 1 | DISK 2 | DISK 3 | DISK 4 | Physical disks | 64K | 64K | 64K | 64K | Stripeblock (RAID-level) | 32K | 32K | 32K | 32K | 32K | 32K | 32K | 32K | Filesystem block (what you set during partitioning and formatting) | RRRRRRRRRRRRRRRRRR | ........................ | R=Read request (71KiB) | As you can see, all filesystem blocks fit into the stripeblock and are perfectly aligned. Now that i'm thinking about it, i think you're referring to the RAID5 write hole. In my example of 71KiB read, there is no problem and no misalignment. But when writing sequentially, some dumb drivers don't buffer the requests. So in fact all I/O has to be done in 2 phases - first read parity, calculate new parity, write parity+data. Besides this being extremely slow it also wears your disks because they will have to seek excessively. But intelligent RAID5 drivers will buffer the writes and always write full stripe blocks (192KiB in this example when using RAID5). Once writes come in from the operating system, it will be stored and if within x seconds a full stripe block is detected it will be written out with full efficiency and speed. That's why the geom_raid5 driver in FreeBSD/FreeNAS manages to achieve 400MB/s write performance in 8-disk RAID5 (disks do 70MB/s). Beating my Areca ARC-1220 by the way, which is still faster for any random IOps task. For some dumb drivers like nForce MediaShield RAID5, it is possible to see high performance to manually correct for the RAID5 write hole, which is i think what you meant. But why should people trust a dumb implementation for their data? If you use Windows, just buy a good controller and a BBU, then you'll be fine. If you use Linux or BSD you can save yourself the cost for such an investment, because the software is smarter. I ofcourse like the last approach, but i understand people want to stick with Windows. Ah he has three drives, sorry i overlooked that This is odd then, but i don't know his controller. Maybe he can try to tweak it, my Areca has several options in the BIOS menu, like write-back, read-ahead, NCQ, etc. or he can use the onboard controller for his raptors in RAID0 which should also be very fast. Just don't use them in RAID5. BTW, nice to see some well informed users related to storage, i was beginning to think they were extinct.
  2. Enlightenment

    Raid 0 performance

    On Windows, all striping RAIDs have a misalignment. Unfortunately Microsoft did not bother to use a proper offset for their partitions. It could easily make sure the partitions starts at exactly 1MiB offset so all stripesizes up to 1MiB (1024KiB) are supported without misalignment. The larget the stripesize, the less performance will be lost to misalignment. I've experimented with stripesizes in excess of 100MB and found no real reason to use any lower stripesize than 128KiB. Bad drivers that use stripes as they were sectors, may spoil performance though. Anyway, ATTO shows you can get 200MB/s write and 175MB/s read which means about double speed. So your RAID0 is functioning as expected.
  3. Enlightenment

    Raid 0 performance

    Try my mirror: http://www.fluffles.net/files/ATTO-256.exe
  4. You want the spindle speed to be lower, for a harddrive that is aimed at mass storage, not really as a system disk. For storing data like MP3's, Video's, etc. Performance is not really important; since the disk is many times faster and you only need x KB/s to listen to music or video; more will not help. What does help is a quiet, cool, energy efficient and stable drive, as in vibrations. So for storage, i prefer 5400rpm to 7200rpm. Especially the new WD Green 10EADS 1TB drive is very nice with 334GB per platter and even lower power consumption (2,9W idle) with a modest price tag. They may also be more reliable than their 7200rpm counterparts because of less friction and less heat generation. And you don't need cooling for them, and they can never overheat themselves in a normal environment. That's a good thing.
  5. Enlightenment

    Best ATX regular case for 18+ hdds?

    If you use 5,25" bays housing multiple 3,5" disks, a cheap choice is the Chieftech BA and CA series, have 8 cooled HDDs on the bottom and alot of 5,25" bays; plus you can rig something above the power supply there is also a cooling hole there. It has two 120mm back fans and tree 92mm front-side fans and sells for about 75 euro.
  6. Enlightenment

    How many drives for raid 0?

    The OCZ Core V2 is said to have corrected this extremely long write latency. Though i do not have clear confirmation, anyone considered buying an SSD is wise to checkout all relevant reviews and comparisons. If this is fixed, it should prove a good value for money, put alot of them in RAID0 on an intel ICHxR with write caching enabled (default is disabled) and the first X MB you write will slide into your RAM so you won't notice any slowdown in regular operation except for large writes but those the SSDs do pretty well. Its the random writes that hurts, because the SSD has restrictions when writing to small parts in the memory, it must first be read, then erased, then re-programmed for a full cell block, so writing 4KB might mean the SSD has to process over 2MB, you can see this is going to hurt performance. Still, SSDs should provide the user with a new experience of instant-response, something that should be very appealing to all regular computer users.
  7. Enlightenment

    The most reliable 1 TiB disk?

    A more energy-saving harddrive like the Western Digital Green 10EADS 1TB should be more reliable, because it cannot overheat with idle usage of just 2,9W. Your drive is about 7-8W and requires cooling.
  8. Enlightenment

    Raid 0 performance

    Where did you connect the drives to, the onboard SATA controller supplied by the chipset? Or another onboard controller, or an add-on controller? IF you're using a PCI controller, you should use the ports supplied by the chipset, those are often the fastest. Also, please test performance with ATTO (the 256MB version) and post it here.
  9. Enlightenment

    Which controller to use?

    Could you test the SiI-3132 with ATTO-256 please? (mirror) I have a few of these PCIe x1 addon controllers with SiI-3132 on it, and it achieves about the same STR (though lower IOps) as my onboard controller; i.e. not capped. But then the disks it is controlling are not as modern as yours. The PCI-express x1 lanes are handled by the southbridge. So you may attempt to put the controller in a PCI-express x16 slot (yes that should work). This way you can rule out the southbridge since these x16 ports are connected directly to the north bridge. But please test with atto; maybe just your benchmark is wrong.
  10. Enlightenment

    Can you interpret these results for me?

    If you write to a bad sector, your harddrive can swap out the bad sector and use a reserve one. Thus, a complete re-write if the harddrive would allow the disks own mechanics to heal itself. I prefer to do this with a linux livecd and execute the command: dd if=/dev/hda of=/dev/hda bs=1M This will copy data from the input file (if is set to /dev/hda) to the output file (set also to /dev/hda) with a blocksize of 1MiB. You need to change /dev/hda to the actual name of your harddrive, that's the only real important part. This procedure will not destroy your data! After you did this, you can safely reboot in windows. Unless the drive is really bad it should have fixed itself by now.
  11. First of all use a filesystem-test program like ATTO (found here: http://www.fluffles.net/files/ATTO-256.exe ) HDTune is not suitable to test RAIDs on, they use a queue depth of only 1; whereas NTFS filesystem will use asynchronous I/O and have multiple queued I/O. If your crappy fake RAID implementation fails to provide you with proper sequential performance with the most simple RAID level; RAID0, try using Software RAID instead; should work like a charm, and tested to do over 500MB/s with 8 disks, on Windows XP SP2: Too bad you can't boot from it.
  12. Enlightenment

    Western Digital 333 Platters

    On the info i've found on the net (techreport.com et al) the different product code is: EACS -> 250GB/platter EADS -> 333GB/platter source: http://techreport.com/articles.x/15769
  13. Enlightenment

    How many drives for raid 0?

    RAID0, like all striping RAID, also increases the IOps, not only the maximum sequential throughput in MB/s. The problem is that on Windows, you end up having a filesystem on the RAID0 volume that is not aligned with the striping blocks. This is due to the way Windows partitions and is not easily corrected. The result is that many of the potential of parallellism goes to waste due to two disks being needed for 1 I/O transaction, whereas you want 1 I/O to be handled by 1 disk so the other disks can take other I/O in the mean time. With a perfect alignment, most often found on Linux and BSD/UNIX families, you can boost Random IOps performance considerably: Single drive concurrency Performance in I/O's per sec. average 1 106 106 107 106 4 106 106 106 106 16 116 116 116 116 32 127 125 126 126 128 151 151 150 150 256 156 156 157 156 geom_stripe (RAID0) with 4 drives - 128KiB stripe concurrency Performance in I/O's per sec. average 1 173 173 173 173 4 270 270 270 270 16 338 338 338 338 32 370 370 370 370 128 444 434 434 437 256 465 465 465 465 Analysis: performance increase ranging from 63% to 300%. But you won't see any of that when using Windows without specifically crafting the filesystem to be perfectly aligned. Aside from that, if the topic starter wants performance for his desktop, i think an SSD drive might be the best choice, given you can justify the cost and settle for limited space to store your OS/Apps/Games on. A nice and kinda cheap SSD is OCZ CoreV2 60GB or 120GB. These have very nice performance as well, just not as good as Intel X25-M SSD which is currently too pricey. You can put SSD's in RAID0 without much risk, since SSD's don't just fail they have a predictable lifespan and even when failed you should still be able to read the data, just not write to it. Just know that RAID0 does not improve gaming loading times alot, and certainly not the framerate. Games are too badly programmed to utilize performance potential, when loading data from disk.
  14. Enlightenment

    RAID 2 or RAID 10 (or 0+1)...

    I was thinking more about: - 1x SSD (starting from 160 euro) OR - 1x Velociraptor 10k rpm 300GB OR - 2x Samsung F1 334GB in RAID0 in your desktop, and as backup a single WD Caviar GP 1TB disk, preferably on another pc on the local network. You don't need a hardware controller for good performance, although a write buffer is always nice. But that can also be achieved via software in the systems RAM. For that you'll need journaling to keep it safe for your data.
  15. Enlightenment

    RAID 2 or RAID 10 (or 0+1)...

    I see nothing wrong in using RAID0, just be sure have a backup. An easier solution would be to keep your RAID0, or replace it by two newer and faster disks in a 2-disk RAID0 configuration, and buy an additional disk to hold a backup. If that disk is put in a PC on the Local Area Network (LAN) and you sync files via (gigabit) ethernet, then you have better protection of data than using mirroring RAID alone. Remember: RAID is not a backup, the RAID itself introduced a layer that can fail. Why not buy yourself a good WD Caviar GP 1TB - the Green Power series only uses half the energy a normal disk does: around 4W versus 8W. It also holds a large amount of storage so the drive would be something you can use well even after 5 years. It's perfect for holding storage but not really great as a desktop system disk. If you want to enhance the speed of your desktop system, the ultimate choice may be to buy an SSD since it's just so much faster than ordinary HDD's. It really depends how much space you need for your installed programs, games and other things you want on your system disk. Otherwise, the space offered by SSDs is not enough for you. The fastest 7200rpm disks right now are those with 333GB platters, such as the Samsung F1 334GB. The new 10k rpm WD Velociraptor 300GB may also suit your needs for a fast system disk, but may be too expensive to run in RAID0 config; but i doubt you'll really need that kind of performance level. My personal choice would be to seperate your data and your system disks, by storing all your personal data on a NAS (dedicated pc serving files over the network) and concentrating on performance on your desktop computer system(s). An SSD would be cool, or velociraptor or a RAID0 array of newest generation 7200rpm disks.