On Wed, 2012-02-08 at 15:31 -0500, Jim Rees wrote:
> J. Bruce Fields wrote:
>
> On Wed, Feb 08, 2012 at 12:49:01PM -0500, Jim Rees wrote:
> > Myklebust, Trond wrote:
> >
> > 10GigE + high latencies is exactly where we're seeing the value. Andy
> > has been working with the high energy physics community doing NFS
> > traffic between the US and CERN...
> >
> > CITI to CERN is just over 120ms. I don't know what it would be from Andy's
> > house. Does he have 10G at home yet?
>
> That still seems short of what you'd need to get a 255MB bandwidth-delay
> product.
>
> I'm just curious what the experiment is here and whether there's a
> possibility the real problem is elsewhere.
>
> In my opinion, any fix that involves allocating multiple parallel data
> streams (rpc slots, tcp connections) is masking the real problem. But it's
> an effective fix.
Who said anything about multiple tcp connections? All the slots do is
allow the server to process more RPC calls in parallel by feeding it
more work. How is that masking a problem?
--
Trond Myklebust
Linux NFS client maintainer
NetApp
Trond.Myklebust@xxxxxxxxxx
www.netapp.com
��.n��������+%������w��{.n�����{��w��� jg��� �����ݢj����G���� ���j:+v���w�m������w�������h�����٥
