• Content Count

  • Joined

  • Last visited

Community Reputation

0 Neutral

About tjoff

  • Rank
  1. Count me in as well. My thought exactly, and I'm getting tired of TrueImage so I've been open to alternatives as of late.
  2. I also contacted Western Digital if they had any success stories they could share but whatever they know they wouldn't share any of it. Though that is understandable, not that easy for a company to comment on other companies products - the intel rep. suggested that I'd give it a try though. Bummer... Still haven't found any reasonable priced (for my purposes) hardware that anyone seems to be able to comment on regarding the >2 TB issues.
  3. So, I mailed LSI and called intel. They guy at intel that I talked too didn't think there would be any problems. Although since they don't have any >2TB drives in their compatibility list he wouldn't comment on any drives and couldn't say more than that he thought it would work (but that any drive not on the list could have compatibility issues). He said that there was a limit on 2 TB logical drives on old controllers, but if that includes whole arrays I guess it's quite an old issue. He said that there, if anything, could be a problem with the onboard raid5 but that logical drives should work. I never really convinced myself that he knew what he talked about, it seemed like he assumed that breaking the 2 TB barrier was the same as any other increase in drive capacity so I'm not convinced that he actually knew anything about the specific issue. LSI (fast mail-support ) wasn't so optimistic: Although that answer seems straightforward enough I kind of believe thats the best and only answer I'm going to get from a manufacturer that doesn't have any >2TB drives on their compatibility list (and those doesn't seem to be updated that very often) irregardless of whether it actually can handle >2TB drives.
  4. Thanks. If the LSI SAS1068E and/or cards based on it's chipset (since intel SASUC81 is about the only 1068 card available here) suppots >2TB drives that would save me some headache But I've had no luck finding that out with google.
  5. Anyone that have read a review of a 3TB drive will be aware that there are quite a few compatibility issues. I have looked but haven't really found any real information about what controllers that support >~2TB drives or if regular controllers are as affected as motherboards seems to be. The WD Green drives come with the HighPoint RocketRAID 620 so I'd guess that it's a safe bet when working with >2TB drives. And I did find this post: But other than that I haven't found much information about it so I thought it would be nice to have a thread dedicated to hardware that does (or doesn't work) with >2TB drives.
  6. Thats my understandning as well. But I have no idea how that have come to be called garbage collection (especially since there are drives that actually do garbage collecting by checking the content of the file system). And, unfortunately, the algorithm won't affect the long term effects of not having TRIM since it can't do anything about it. Of course a good algorithm will handle the smaller pool of unused blocks better, but there is no way for it to avoid the available blocks to get used up.
  7. Without TRIM it doesn't matter whether the "garbage collection" is aggressive. The drive knows nothing about the data it stores and if the drive gets filled the best thing the garbage collection can do is just rearrange the data but it can't free data it don't know anything about. I'm not sure why the term garbage collection is used. It is quite rare that the drive actually snoops the traffic and sees what the filesystem is doing and makes decisions based on that.
  8. I don't get your point. They say that it's a fine tuned balance. Not that the spindle speed is varying. And they have, to my knowledge, never mentioned that it is. Back in the day when the green drives where new and this was a common belief I really doubted it and searched for any claim where WD had said this and didn't find any. What they did was not to specify exactly which spindle speed it used. This was obviously since they would have gotten a bad reputation for being about the only modern 3.5" drive (at the time) with less than 7200 RPM. This has already been stated in the thread... Yes thats a shame. I wouldn't say intentionally mislead though. I'd guess that they do this so that they can phase out the old drives and replace them with new more easily. But anyway, I agree. That kind of information should be easy to obtain. Probably? In your first post you claim it as a fact. Uncool. Got anything to back it up with? Is there even a standard ATA command that does what WDTLER.EXE does? As does all other manufacturers of 4KB drives... You seem to have something against WD. Instead of focusing on one brand you could extend your rant to include basicly all other manufacturers as well, but that wouldn't be as fun or what?
  9. tjoff


    Oh, don't know why but I assumed that the timout-settings only was relevant for drives used within a hardware-raid array.
  10. tjoff


    Assuming you don't use ZFS or any other form of software raid, right? I haven't seen any information about whether software raid can cope with this.
  11. Stores here in Sweden list it as being available from the 13th of November. The sad part seems to be that you can't get it without the HBA? I'm a bit tempted in putting a couple of theese in raid, but then I will end up buying a bunch of HBAs that I don't have any need for... I haven't seen any information about whether the HBA bundled retail version is the only one that will be sold for now? I'm assuming thats the case since the price of the 3TB WD is over 20% more than the external Seagate drive. And with Seagate you get an external chassi and external power and a USB->sata converter which should cost about the same or more as the bundled HBA so the price is a bit disappointing even considering it's the first internal 3 TB drive and that it includes an HBA. Or I should perhaps say that Seagates drive is cheap rather than WD being expensive in this case. Still, the HBA part is annoying.
  12. LOFI doesn't seem like a sensible solution for a large full raid encryption. I could be wrong but it doesn't seem to have the greatest reputation for other things than small solutions. Performance being one issue. But I could be wrong, also doesn't seem like a widely used solution for my purpose so I'm a bit skeptical.
  13. I didn't find the LSI SAS1068E anywhere close to where I live. I found a supermicro PCI-E 8x card based on the LSI 1060E chipset, but it's an UIO version and it cost about the same as the intel-card. The intel card is also based on the LSI 1060E chipset so I guess it's pretty much the same thing anyway(?). As for ZFS expansion vs. BTRFS you are probably right. I've been kind of turned off by the whole solaris thing since Oracle took over, and I've searched a bit but haven't really found out if raid-z expansion really is something that Sun/Oracle are even interested in. The best I've read is that people are talking about a coming feature that will make expansion quite trivial to implement but could just be hearsay. Hmm, and I thought zfs-crypto was considered stable and included in OpenSolaris already :| That probably kills it for me in the short term. When the whole solaris+oracle(+illumos?) thing have calmed down and when raid-z expansion and ZFS-crypto are available I'll consider moving. Until then I guess that mdadm raid6 + dmcrypt + EXT4 will have to suffice. superlgn: Thanks for looking it up. Finding miniSAS => 4 angled SATA cables have proven to be quite difficult and a surprising show-stopper. Don't really know how to proceed with out them. Found a few more: But they are too short... I could use angeled SATA adapters ( ) but it would be quite expensive and I'm afraid about glitchy connectors as it is and it probably won't do wonders with the signal quality either (but perhaps thats not an issue).
  14. tjoff

    Intel SSD RAID Review

    Not a single word about TRIM? (didn't read through everything, apologies if I missed something) If it hasn't changed since last time I checked there is no card that can do TRIM on RAID array. Having no TRIM at all can, in time, easily outweigh any benefits of RAID and thus perform much slower than even a single drive. You could probably do some hacks to improve non-TRIM performance such as ensuring that some (TRIMmed) space will never be used - i.e. before making the array, make sure that everything is TRIMmed and leave x % of the array unused at all times (by making sure that you don't even partition x % of the array for instance). That will save some space for the controller to juggle data around and should improve write speeds significantly when the array gets full. But a better solution in many cases is simply to use the two drives as is. For instance one for the OS and one for applications. Especially with the newer and faster SSDs coming, pure bandwidth isn't the whole picture and with Intel G3 around the corner the use of SSD-raid should, for most people, not be that appealing. But really, I'd recommend a single large SSD for most of the cases. A 160 GB drive is often more or less two 80 GB drives in the same package anyway so you could say that you already are running them in RAID 0 in some sense (other bottlenecks take place (the controller for one) but write speed for the G2 160 GB is much better than the write speed for the G2 80 GB). And thats probably the main reason for why the budget-drives are smaller and why the smallest G3-M drive will be 160 GB and not 80 GB and why there is no X25-M 40 GB. They can't make smaller drives in the same performance-league as the larger ones. Please correct me if I'm wrong, I might be a bit outdated on the subject but I use full disk encryption on my G2 80 GB so I sure now how devastating the lack of TRIM can be for performance when the drive gets full.