-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi,

It is exactly that option.  With some small changes.  The draft has a
fixed time HAMMER_TIME.  Unbound has a hard-coded percentage, 10%,
which is not configurable (by choice, no option that hurts).  This
setting causes an upper bound on the extra work of about 10% and about
10% more traffic towards nameservers (so the prefetch option is not
deleterious to others).  Authority server owners can influence their
prefetch behaviour via the TTL (a larger TTL has a larger
'hammer-time' of 10% of TTL, but also still a larger cache-TTL as
well, and also the relative popularity that is necessary to be
prefetched goes down as you only need queries inside of a longer time
period to be prefetched).

We tested this in a large production resolver and found that the
benefits are small.  Likely when the cache expires there is no more
spike in latency for a couple of customers.  The cache hits increase
modestly, but not spectacularly; some percents more cache hits.

In the draft the "touch this" phrase is very funny, however I would
prefer implementation-independent phrasing, i.e. 'do not perform
prefetch for this element'.

Is a draft necessary for this?

Best regards,
   Wouter

On 07/02/2013 07:58 AM, Matthijs Mekking wrote:
> This sounds a lot like prefetch in unbound, and the configuration
> option gives some analysis on increased traffic.
> 
> prefetch: <yes or no> If yes, message cache elements are prefetched
> before they expire to  keep  the  cache  up to date.  Default is
> no.  Turning it on gives about 10 percent more traffic and load on
> the machine, but popular items do not expire from the cache.
> 
> 
> Also, if the original TTL of the RR is less than STOP * HAMMER_TIME
> then the cache entry, it cannot be used anymore and the resolver
> should "Break it down".
> 
> Best regards, Matthijs
> 
> On 07/02/2013 03:44 AM, John Levine wrote:
>>> We would like to draw your attention to a new draft.
>> 
>> It looks like it should work, assuming that your cache uses the 
>> existing logic to remember queries in progress so it doesn't
>> hammer a record that's already in the process of being
>> refetched.
>> 
>> My main observation is that I have no idea what the tradeoffs
>> are between the increased DNS traffic and faster responses to
>> clients. Have you done simulations or live experiments?
>> 
>> R's, John _______________________________________________ DNSOP
>> mailing list [email protected] 
>> https://www.ietf.org/mailman/listinfo/dnsop
>> 
> 
> _______________________________________________ DNSOP mailing list 
> [email protected] https://www.ietf.org/mailman/listinfo/dnsop
> 

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.13 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/
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=dXxa
-----END PGP SIGNATURE-----
_______________________________________________
DNSOP mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dnsop

Reply via email to