David Haller wrote: > Hello, > > On Fri, 22 May 2020, antlists wrote: >> On 22/05/2020 19:23, Rich Freeman wrote: >>> A big problem with drive-managed SMR is that it basically has to >>> assume the OS is dumb, which means most writes are in-place with no >>> trims, assuming the drive even supports trim. >> I think the problem with the current WD Reds is, in part, that the ATA-4 spec >> is required to support trim, but the ATA-3 spec is the current version. >> Whoops ... > ATA-8 is the current spec. Though practically unused. The used spec is > ATA-7 in virtually all drives for IIRC the last 10ish years or so. > > Did you mean SATA specs? Well, then there's only SATA-1 (1.5GBit/s), > SATA-2, (3.0GBit/s) and SATA-3 (6.0GBit/s), and of the latter SATA > revision 3.1 introduced TRIM[2]. Oh, and rev. 3.3 introduced some > extras for SMR [3]. > > HTH, > -dnh > > [1] https://en.wikipedia.org/wiki/Parallel_ATA > [2] > https://en.wikipedia.org/wiki/Serial_ATA#SATA_revision_3.0_(6_Gbit/s,_600_MB/s,_Serial_ATA-600) > [3] https://en.wikipedia.org/wiki/Serial_ATA#SATA_revision_3.3 >
Given the info in link 3, does that mean that OSs and their layers will start adding options to help make SMR drives work better? In other words, file system support, LVM support if needed and maybe even something that will make RAID setups work better? From the way it sounds, right now there is basically none of that. Everything happens on the drive and once it hits the limit of what it can handle, things start screeching to a halt with the OS left in the dark about what is going on exactly. If SMR is going to be the new thing, it needs to work better, RAID seems to really need that help. Dale :-) :-)
