On 4 December 2015 at 01:25, lejeczek <[log in to unmask]> wrote:
> hi,
> from the reading I've done there is that first conclusion - it may different
> a lot with each FS - people theorize. I wonder if anybody put them into
> practice.
> I wondered if btrfs/xfs was any good, etx4 in current kernels does not
> support "packed_meta_blocks" yet. Some hands-on experience.
> thanks
>
I don't think most of the people who actively post on this list are
into buying cutting edge hardware but are usually dealing with 2-5+
year old hardware. That said SMR should be treated more like a (Write
Once, Read Many) WORM drive. If you are rewriting data to it.. it is
going to be abysmal in speed and use.
http://www.storagereview.com/seagate_archive_hdd_review_8tb
Write performance on SMR seems to be 1/20th of PMR drives so it isn't
really going to matter what the filesystem underneath is because the
largest variable is the drive itself. [EG zfs might increase by 20%
but not the 2000% needed to match PMR]
--
Stephen J Smoogen.