Hi Alfredo,
That is very helpful.
We are looking at virtualising Ntopng Edge using KVM on a test Dell server
with Intel E5-2640(2.50 Ghz, 6-core) and 96GB of RAM.
Question 1 - I assume your 600Mhz test was bare-metal, correct?
Question 2 - Would we be looking at numbers quite similar to this? (I
assume there is some kind of ceiling for performance).
Question 3 - Which upgrade would have the biggest improvement on latency?
(e.g. more cores, faster clock-speed, faster SSDs, more RAM etc.)
Question 4 - Are there any caveats with virtualising Ntopng Edge with KVM?
Is this expected to affect performance, and if so, any guidelines to
mitigate it?
Question 5 - How can we measure our own latency? Does Ntopng Edge have some
kind of diagnostics mode, or debug page which presents this data?
Thanks,
Victor
On Fri, Aug 17, 2018 at 5:34 PM Alfredo Cardigliano <cardigliano@ntop.org>
wrote:
> Hi Victor
> on a low-end embedded system (600Mhz) we measured an avg
> added latency of ~0.15 msec, and a max of ~1.75 msec, with
> respect to a clean linux bridge.
>
> Regards
> Alfredo
>
> > On 15 Aug 2018, at 23:41, Victor Hooi <victorhooi@yahoo.com> wrote:
> >
> > Hi,
> >
> > I'm curious what sort of added latency we can expect an inline Ntopng
> Edge server?
> >
> > Assuming that the machine is reasonably well spec-ed - e.g. 16GB of RAM
> (or more), 4 or more cores (I assume Ntopng Edge is multithreaded, right?),
> modern SSDs - is there some baseline figure (or range) in milliseconds that
> it will always add, no matter what?
> >
> > (Either from Ntop testing, or perhaps people's experience in the fields)
> >
> > Regards,
> > Victor
> > _______________________________________________
> > Ntop mailing list
> > Ntop@listgateway.unipi.it
> > http://listgateway.unipi.it/mailman/listinfo/ntop
>
> _______________________________________________
> Ntop mailing list
> Ntop@listgateway.unipi.it
> http://listgateway.unipi.it/mailman/listinfo/ntop
