Re: [Qemu-devel] [PATCH] sheepdog: implement direct write semantics

[Paolo Bonzini]

Il 09/01/2013 11:36, Liu Yuan ha scritto:
>> But why is it useful to force-disable writeback caching?  Do you have
>> > any performance numbers?
> This is not related to performance. What I want is to allow users to
> choose Sheepdog cache mode (writethrough/writeback/directio) from cache=xxx.
> 
> Obviously writeback mode show much better write performance over
> writethrough cache in Sheepdog. NOTE: writethrough cache in Sheepdog is
> a readonly cache which is always consistent with cluster data thus need
> no flush at all. This is to boost read performance for read intensive
> Guest over Sheepdog iamges.

Ok, so the questions are (compared to older QEMU's writethrough mode):

1) how slower is QEMU's emulated-writethrough mode for writes, due to
the extra requests?

2) how slower is QEMU's writeback mode for reads, due to the different
structure of the cache?

Paolo