[Sheepdog] Databases, write barriers etc

[Brian Candler]

On Tue, Aug 09, 2011 at 01:18:57AM +0900, MORITA Kazutaka wrote:
> I think the performance could be a problem.  The latency of a Sheepdog
> is higher than other local disk formats because of data replication
> and network latency, so Sheepdog would show a poorer performance with
> something like databases.

If an app wants to update a single block, will this require Sheepdog to
fetch an entire 4MB object, modify part of it, and write it back?

You're right, I need to do some performance testing.

Thanks,

Brian.