Defiler

Patron
  • Content Count

    585
  • Joined

  • Last visited

Community Reputation

0 Neutral

About Defiler

  • Rank
    <a href='/patron.html'><b>StorageReview Patron</b></a>

Profile Information

  • Location
    Florida
  1. Defiler

    OCZ's 1TB SSD

    Huh; is SAS fast enough to handle those transfer rates? Cool. There are cheaper ways to get that many iops if you have room for more devices though.
  2. Defiler

    OCZ's 1TB SSD

    Will be interesting to see what the 256GB model goes for. That is a pretty useful size, and there's a lot to be said for the simplicity of non-RAID setups.
  3. That is actually 100% awesome and I am going to do that immediately.
  4. Defiler

    64-bit OS?

    I suspect what he was talking about was the 'No Execute' feature of x86_64 CPUs in native mode: http://en.wikipedia.org/wiki/NX_bit In particular check out the list of supported hardware under the 'Microsoft Windows' section.
  5. Hopefully this guy has upgraded since then. I read the first post of the thread and thought "oh god, why would you do that?", then I saw the date on it. Haha.
  6. Defiler

    What would you do

    Do you have some spare disks available? You can then image them individually and try out various options using that spare controller. Do you have other replicas of this data, or do you actually need to recover it? Hopefully you have spares, but if not, this is the best thing I've found: http://www.r-studio.com/ To quote from the (sometimes oddly worded) product manual: "If, due to hardware failure, a hardware volume set or RAID cannot be accessed, or due to data loss your system does not recognize a software volume set or RAID, and you know what hard drives were in it, you may create a Virtual volume set or RAID and process it like a real software volume set or RAID or software volume set or RAID." [PREACHY]In my opinion, it isn't worth storing data unless you are storing it at least twice, preferably simultaneously.[/PREACHY]
  7. Defiler

    Ram Disks vs SSD vs Hard Disk

    Even then, there are generally better ways to do that that manually creating a ramdisk (Varnish, memcached, Cassandra, etc, etc). The basic problem with ramdisks is that they have filesystems on them. (Though of course everything Brian said is also true) Operating systems already have lots of good tools for storing application data in memory, and the filesystem, block storage metaphor, etc, all just get in the way. 95% of the time, it is better to use that RAM as a cache than to put a filesystem on it.
  8. Now I want an edible USB drive, so that this would actually work.
  9. Actually, I have what I need to perform tests for this set up right now. Intel X58 board, two SSDs that support TRIM, etc. That being said, I can't think of a sane way to test this; these are SandForce drives, so they work fine even when you don't send TRIM. Anyone have any ideas? I suppose it would be best to try this with X25-M drives, but I don't have any.
  10. Defiler

    Best EHD's for DVR Storage

    You want two things; 1) A well-made enclosure that will keep your disk cool in a cramped space. 2) A hard disk inside it that is expecting to be on at all times. What I would do (sadly not the cheapest path) would be to get a single-drive Icydock enclosure: http://www.icydock.com/externalsata.html ..and then something like this: http://wdc.com/en/products/Products.asp?DriveID=615 Most of the major HDD companies make DVR/streaming-oriented low-power 3.5" drives these days. You may want to check Newegg or Amazon to see which ones people like the best. You can also use a 'enterprise-grade' 3.5" SATA drives as well, but they usually aren't worth the money for this kind of thing.
  11. One disk in CrystalDiskMark: RAID-1 in CrystalDiskMark: Unfortunately CDM was very unhappy with RAID-0, at least with Windows 7 soft RAID. The one time I waited long enough to see some results from it, it was getting 3.5MB/sec sequential. Since all the other benchmarks worked fine, I'm assuming this is a CDM problem. Personally, I prefer AS-SSD to Crystal. One disk in AS-SSD: RAID-0 in AS-SSD: RAID-1 in AS-SSD: Test System: Intel i7-920 6GB DDR3 Corsair Dominator ASUS P6T6 WS Revolution ATI Radeon 5870 (not that it matters) Boots from a 1TB Caviar Black (-FALS model) Windows 7 Professional x86_64
  12. Wow these things are quick. I will post some results tomorrow; ran various combinations in Crystal and AS-SSD. I wish I understood why random reads are slower than random writes. For fun, I pulled out an Mtron SLC drive as well (MSP-SATA7025-032) to compare against the Jan 2008 state of the art.
  13. OK, cool. Same version I have installed. Can you go into a little more detail about the new Iometer patterns? The 2008-06 version doesn't come with any 2MB, random, or 'named' access patterns (e.g. the old 'file server' pattern). The Testbed4 page doesn't say precisely what the settings are, and the new reviews don't use exactly the same patterns.
  14. Are the StorageMark 2010 traces available somewhere? Also, are you still using Iometer v2004.7.30, or a newer build? Got these things installed now, but just spent a while hunting around the site for details of the new tests. As far as I can tell, the Kingston V+ review was where the 2010 scripts debuted?
  15. In the final config, one is just going to be a spare in case I need to RMA the other. For testing, I was planning to try them in software RAID 0 and 1 just to see how it behaves.