Troy Benjegerdes wrote:
This kind of a claim needs some benchmark data to document it.
We've implemented both for our vdisk driver on 390. At least on our
platform, merging in the host is preferable because vmenter/vmexit is
very fast and we would merge twice because we submit the result via
io_submit() system call from host userland.
In the end you'd need to measure on all platforms this is gonna run
on, if you want competitive benchmarks. That makes the benchmark
argument somewhat fuzzy.
I'll make the counterclaim that you *should* be doing I/O scheduling in
the guest, both to be able to test new I/O schedulers, and to provide
a set of pre-scheduled I/Os so the host has to do less work.
What makes the work to merge requests less, if done in the host? Is'nt
it the same code really if you run Linux in both host and guest?
If someone really needs the host to be doing I/O scheduling, and not the
guest, then why are they using virtualization in the first place?
Just to run multiple operating systems on one box. Funny question...
so long,
Carsten
_______________________________________________
Virtualization mailing list
Virtualization@xxxxxxxxxxxxxxxxxxxxxxxxxx
https://lists.linux-foundation.org/mailman/listinfo/virtualization
