Brian,

[ sorry for coming back to this thread so late ]

On Wed, 6 Nov 2013 17:02:50 -0800
Brian Somers <[email protected]> wrote:

> Using a sample of our US traffic over a 7 minute period, I see
> 112,044 lookups of www.facebook.com out of a total of 4,081,834
> queries (2.74%). That's 266 www.facebook.com queries per second.
> 
> Doing a lookup where our resolver is only missing
> star.c10r.facebook.com (which the www.facebook.com CNAME points at;
> the record with a 60 second TTL) takes 1.6ms. Doing a lookup where
> everything's cached, our resolver takes 0.7ms.
> 
> Given that our prefetch implementation actually charges the client
> that first queries the record in the last 3 seconds of its TTL, and
> given that we now cap our resolvers at 6 threads, this means that in
> the 7 minute period, we had 7 clients waiting for the upstream rather
> than 265 times 7 cilents.
> 
> This is a relatively big win, but is also a drop in the ocean -- not
> much more than a second of latency shared among these clients.
> 
> The real win for us is that we don't ask facebook the same question 6
> times per minute (one per thread).  Instead, we ask once every 57
> seconds.  This is bigger in the grand scheme of things because it
> makes the facebook auth servers happier :)

First, thank you for providing actual data! Much appreciated. :)

But... while the Facebook case is interesting, I am thinking you'll get
a lot more win from sites with long latencies. So, looking at #5 on the
top 1 million web sites list: baidu.com. From here I see latencies
between 300 and 500 msec. The TTL for the A record of baidu.com is 600
seconds.

It's not just China. Going to #16 on the list is wordpress.com, which
gives me times of 100 to 200 msec from all of its 6 (!!!) name servers.
At #29 ask.com has times between 90 and 200 msec. At #37 apple.com also
has times between 100 and 200 msec.

(I didn't check all of the names in between, since I expect that
yahoo.co.jp and yandex.ru and so on probably only care about "local"
traffic in the same way that Baidu seems to. I took special care to
single out Apple because I'm mean that way.)

The busy server + long latency is where we'll see the most win, IMHO.

Cheers,

--
Shane

p.s. What's the point of having a ton of name servers if they're all in
     California/Beijing/wherever? People, if you're going to have a lot
     of NS for your domain, please peer the network of at least one
     server in AMS-IX so I can get fast DNS resolution from here in
     Holland. Thanks! ;)
_______________________________________________
DNSOP mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dnsop

Reply via email to