enetec Posted June 7, 2015 Share Posted June 7, 2015 Since what I've understood, setting up a cache pool, unRAID make use of one of Btrfs features to make a 'pseudo' RAID1 between used disks. What it's still not clear to me is how much this differs from standard RAID1 and how does it match different capacity disks... What's the total capacity obtained in this case? More... What's the point on using a cache pool of 4 identical drives? There is something not clear to me and my searches have not returned usefull results.... Can someone explain me better how this feature works technically in details? Link to comment
trurl Posted June 7, 2015 Share Posted June 7, 2015 I haven't really bothered to study it in detail but I think you get half the total, with the data stored in such a way that any data stored on one drive has a duplicate on another drive. This duplication is probably not at the file level. I could be totally off-base though. You have probably already seen the calculator. Link to comment
garycase Posted June 7, 2015 Share Posted June 7, 2015 You can have 2, 3, or 4 drives in your "pool" and it will work just fine. The capacity will, as trurl noted, be 1/2 of the total space of the drives, and files will be written in such a way that every file is on two DIFFERENT disks ... so the pool is fault tolerant. i.e. if you have 3 256GB SSDs, you'll have 384GB of protected cache. Link to comment
garycase Posted June 7, 2015 Share Posted June 7, 2015 ... in fact, here's a recent post where this topic was discussed a bit ... and it shows a pool of 3 128GB drives with the resulting 192GB of protected cache: http://lime-technology.com/forum/index.php?topic=40410.msg381272#msg381272 Link to comment
enetec Posted June 7, 2015 Author Share Posted June 7, 2015 Ok, perfect, but if working this way, what's the point (as I saw somebody has posted it...) in making a pool of e.g. 1 TB drive + 1 250 GB drive? He will get only 250 GB of usable cache pool and most of the bigger drive will be unused...! Am I wrong somewhere? Link to comment
trurl Posted June 7, 2015 Share Posted June 7, 2015 Ok, perfect, but if working this way, what's the point (as I saw somebody has posted it...) in making a pool of e.g. 1 TB drive + 1 250 GB drive? He will get only 250 GB of usable cache pool and most of the bigger drive will be unused...! Am I wrong somewhere? Didn't notice that post. What you say makes sense to me and the calculator agrees. Link to comment
garycase Posted June 7, 2015 Share Posted June 7, 2015 You're correct => clearly you can't provide protection for a drive that's larger than all the other drives in the pool combined. He could make it an unprotected pool, and then all of the space would be available ... but that kinda defeats the whole idea of the pooling concept. Link to comment
enetec Posted June 7, 2015 Author Share Posted June 7, 2015 Ahahahah... it seems to me that I'm not the only with confuses ideas about cache pool!! Even if I'm not a lot interested on using it, I'm very interested on how it works exactly technically (it's the same which attracted me to unRAID... see exactly how it worked. Fine I can say now! ). From what I've understood: - two drives: space available should be equal to smallest one as in "standard" RAID1 (or a "degenerated" RAID5 if you prefer... - after you'll understand why I call it this way - with one data and one parity disk. Using even or odd parity we can have a real mirror or "negative" one) - three drives: it seems that even in this case, space available should be equal to smallest one, creating a "double parity RAID1" (or a "degenerated" RAID6 if you prefer, with one data and two parity disk. Same consideration as above on parity) - four drives: this is simple if all drives are equal: space available should be equal to the sum of two drives (spanned) and then mirrored on other two. This way it seems about a RAID10 but spanning instead of striping (similar to unRAID data disks that are about a RAID4 but without striping too). BUT with 4 different drives? I've some ipothesys but I'm not sure on either of them. It could be smart and use more space or quite "dumb" and be limited by smallest drive again... Anyway, one thing is clear to me: most are using "strange" cache pool without well understanding how they will be used by system... and this could be potentially Dangerous IMHO... :'( Now... Tom/Limetech will come and will say I'm wrong all the way! Link to comment
Encino Stan Posted June 8, 2015 Share Posted June 8, 2015 - three drives: it seems that even in this case, space available should be equal to smallest one, creating a "double parity RAID1" (or a "degenerated" RAID6 if you prefer, with one data and two parity disk. Same consideration as above on parity) For 3 drives you could have a configuration of 128 + 128 + 256 = 256. Each of the 128 drives has its data duplicated on the 256 drive. Link to comment
trurl Posted June 8, 2015 Share Posted June 8, 2015 - three drives: it seems that even in this case, space available should be equal to smallest one, creating a "double parity RAID1" (or a "degenerated" RAID6 if you prefer, with one data and two parity disk. Same consideration as above on parity) For 3 drives you could have a configuration of 128 + 128 + 256 = 256. Each of the 128 drives has its data duplicated on the 256 drive. Yes, that is what the calculator says also. Link to comment
enetec Posted June 8, 2015 Author Share Posted June 8, 2015 - three drives: it seems that even in this case, space available should be equal to smallest one, creating a "double parity RAID1" (or a "degenerated" RAID6 if you prefer, with one data and two parity disk. Same consideration as above on parity) For 3 drives you could have a configuration of 128 + 128 + 256 = 256. Each of the 128 drives has its data duplicated on the 256 drive. Yes, but I don't think this is selectable at the moment... Am I wrong? Link to comment
Recommended Posts
Archived
This topic is now archived and is closed to further replies.