Hey Matthijs,

On 15 August 2016 at 06:32, Matthijs Mekking <matthijs@pletterpet.nl> wrote:

Hi Jan,

Thanks for your response. Some comments inline:

On 15-08-16 14:29, Jan Včelák wrote:

Hi Matthijs,

processing of queries in Knot DNS is synchronous. So the UDP thread is
blocked until the query processing is finished. This usually doesn't
matter for authoritative server because the server can construct the
response immediately.

For dnsproxy, this is a problem. The dnsproxy module establishes a TCP
connection to the resolver and waits for the answer. During that time
the UDP handler thread is just idling.

The dnsproxy AFAICS will use the same transport protocol as the incoming query. In our tests this is all UDP. So I don't think any TCP connection is established.

Also, the query is forwarded to an authoritative server nearby, not a resolver, which should be a lot faster. There is still idle time of course, and perhaps too much.

Even if it's nearby, it's the latency that kills it. I think the number of workers in Knot is ~ number of cores ~ number of proxied requests in-flight. Resolvers, proxies and LBs use coroutines, event loops or pipelines to resolve queries asynchronously, so they are able to have three orders of magnitude more requests in-flight. There was someone working on reimplementing the proxy module with per-worker libuv loops a few months ago, I don't know what's the status now, but that would help a lot.

I'm affraid this couldn't be fixed easily without deep changes in the
knotd architecture.

Anyway, we are interested in your discoveries.

Me too.

Best regards,
Matthijs

Perhaps you can try increasing the worker count to something like 100? It should help. Honestly, if you're looking for public-facing DNS proxy with filtering and decent performance, you'll be better served by something else. RRL (or rate limits for uncached responses) is relatively easy to implement in DNS firewall in Knot Resolver (have a look at the 1.1 release post) compared to reimplementing the DNS proxy module in authoritative.

Marek

Cheers,

Jan

On 15.8.2016 14:12, Matthijs Mekking wrote:

Hi,

I recently tested the mod-dnsproxy performance and I am disappointed in
the results:

Knot in our test setup can do ~320K QPS.

When using our own proxy in front of knot, we achieve quite a
performance hit, only able to do ~120K QPS.

However, when configuring knot to use the mod-dnsproxy, the performance
drops to ~7K QPS.

I am planning to investigate what causes this significant drop, but if
you have any insights or other measurements already I would love to hear
about them.

Best regards,
Matthijs
_______________________________________________
knot-dns-users mailing list
knot-dns-users@lists.nic.cz
https://lists.nic.cz/cgi-bin/mailman/listinfo/knot-dns-users

_______________________________________________
knot-dns-users mailing list
knot-dns-users@lists.nic.cz
https://lists.nic.cz/cgi-bin/mailman/listinfo/knot-dns-users

_______________________________________________
knot-dns-users mailing list
knot-dns-users@lists.nic.cz
https://lists.nic.cz/cgi-bin/mailman/listinfo/knot-dns-users