On Wed, Dec 23, 2015 at 12:20 PM, Goffredo Baroncelli <kreijack@xxxxxxxxx> wrote: > On 2015-12-23 16:53, Donald Pearson wrote: > [...] >> >> Additionally real Raid10 will run circles around what BTRFS is doing >> in terms of performance. In the 20 drive array you're striping across >> 10 drives, in BTRFS right now you're striping across 2 no matter what. >> So not only do I lose in terms of resilience I lose in terms of >> performance. I assume that N-way-mirroring used with BTRFS Raid10 >> will also increase the stripe width so that will level out the >> performance but you're always going to be short a drive for equal >> resilience. > > In case of RAID10,on the best of my knowledge, BTRFS allocate each CHUNK across *all* the available devices. It uses the usual RAID0 (==striping) over a RAID1 (mirroring). > > What you are describing is the BTRFS RAID1; i.e. LINEAR over a RAID1:each chunk is allocated in *two*, only *two* different disks from the disks pool; the disks are the ones with the largest free space. Each chunk may be allocated on a different *pair* of disks. > Okay so however the chunk is divided up, 2 copies of each chunk division is written somewhere. So I misunderstood, thanks for clearing it up! >> And finally the elephant in the room that comes with the necessary >> 11-way mirroring is that the usable capacity of that 20 drive array. >> Remember, pea brain so my math may be wrong in application and >> calculation but if it's made of 1T drives for 20T raw, there is only >> 1.82T usable (20 / 11) and if I'm completely off in that figure the >> point is still that such a high level of mirroring is going to >> excessively consume drive space. > > Ducan talked about a N-way mirroring, where each disks contains a copy of the same data. Nobody talked about N-way mirroring where N is less than the number of the available disks. > Well that was certainly implied as the unimplemented solution to dropping half the drives that the OP tested. N-way mirroring where N = the number of drives is just Raid1 on crack and not the Raid10 use-case that the OP is asking about. > To be honest in the past appeared some patches to implement a generalized RAID-NxM raid, where N are the total disk, M are the redundancy disks: i.e. the filesystem could allow a drop of M disks (see http://www.spinics.net/lists/linux-btrfs/msg29245.html). > > BR > G.Baroncelli > > > -- > gpg @keyserver.linux.it: Goffredo Baroncelli <kreijackATinwind.it> > Key fingerprint BBF5 1610 0B64 DAC6 5F7D 17B2 0EDA 9B37 8B82 E0B5 Yeah that whole thing is pretty upsetting. -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
