[ovs-discuss] Limitations of Mininet

[Adil Ahmad]

Alright, thank you for the insights.

Adil.

On Tue, 7 Jul 2015 at 2:30 am Ben Pfaff <blp at nicira.com> wrote:

> On Mon, Jul 06, 2015 at 11:28:09AM -1000, Nicholas Bastin wrote:
> > On Mon, Jul 6, 2015 at 3:46 AM, Adil Ahmad <adilahmad17 at gmail.com>
> wrote:
> >
> > > I was trying to get flow statistics and the implementation that I'm
> > > working on requires that I get flow stats in RTT timescales. But, the
> flow
> > > statistics show that the packet count counter is updated after almost a
> > > second for OF 1.0 swtiches using POX. I want the packet counters to be
> > > updated at every iteration around 1 ms. Is that a limitation with
> mininet
> > > or OF 1.0?
> > >
> >
> > This is a limitation of mininet, although what mininet does (and OVS, by
> > extension) is much closer to what you want than any other OF device will
> do.
> >
> > There is exactly zero chance that a real OF network (or any network) can
> > update packet counters on a per-ms basis.  Most hardware does so on a
> 1-10s
> > interval, and the shorter the interval the more sampling is used (and
> thus
> > the more inaccurate it may be).
>
> As long as OVS has mentioned mentioned anyway, OVS itself only updates
> its counters about every second in normal operation.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://openvswitch.org/pipermail/ovs-discuss/attachments/20150706/727d3700/attachment-0002.html>