Hi,
Through my application, I have reached similar figures mentioned by Daniel
in terms of number of queries per second. I am using only one client. For
this rate, I observed some of the queries (1 out of 100000 queries made)
returning null entries to my application. I am using local caching dns
server and all the queries made are for same dns lookup, so I know there is
nothing wrong with my dns server sending null entries. I have also not
ported c-ares yet. I am still using ares-1.1.1 which is the old base for
c-ares-1.4.0 If anybody have observed this problem in ares-1.1.1 please let
me know.
Thanks,
Chirag.
On 9/15/07, Eino Tuominen <eino_at_utu.fi> wrote:
>
> Daniel Stenberg wrote:
> >
> > I'm not aware of anyone having done such performance measurements on
> > c-ares. If you end up doing some, it'd be very interesting to get to
> > know the results!
>
> I have used c-ares as a resolving library for my greylisting server
> Gross. I did once some benchmarking with it, I quote myself:
>
> > FYI, I just run some very non-scientific performance tests. grossd
> > was running on a recently bought low end PC, and clients were run on
> > the same server.
> >
> > 1 client doing 10000 queries: 10 seconds
> > 5 clients doing 10000 queries each: 11.5 seconds
> > 10 clients doing 10000 queries each: 13.5 seconds
> >
> > The last run means 100000 queries running a total of 500000
> > dns-queries, as check-dns was enabled (of course, they were on the
> > same ip address, which means that the results were cached on the
> > resolving servers). With 10 parallel clients you get at least some
> > 650 million messages a day, if your dns servers will cope the load.
> >
> > Rather impressing, mildly put.
>
> --
> Eino Tuominen
>
>
>
Received on 2007-09-17