On 03/04/12 14:28, Steven Whitehouse wrote:
Spinning disks are slow to seek, large arrays even more so.
Large arrays should be much faster, provided the data is in cache.
Or not, when there's a lot of random IO involved and it's not in cache.
I'm talking about arrays such as nexsan atabeasts (a drawer full of sata
drives)
I can't see any mention that bcache supports clusters at all. I don't
think that it is likely to work. Certainly the web page I found suggests
that it doesn't support barriers (silently dropped)
It doesn't and there are specific warnings to disable barriers on ext4
and friends when using it.
Bcache is writethrough by default. Writeback can be enabled but is beta
quality and I think it would conflict badly with clustered filesystems.
What do you mean by flashcache? This perhaps:
Facebook's caching implementation which is almost like bcache but much
simpler in its implementation.
http://www.netapp.com/uk/products/storage-systems/flash-cache/
It looks like a hardware implementation of the same thing, and I can't
see anything to suggest that it is cluster aware on a first reading of
the docs,
There are a few SAN-level accelerators but the cost of those things
starts around $20,000 and climbs from there.
--
Linux-cluster mailing list
Linux-cluster@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/linux-cluster