[Pdns-users] pdns-recursor peering

[Duane]

Brendan Oakley wrote:
> Hi Mick,
> 
> Depending on the application and purpose for wanting a shared cache, I

There is another option that might work here. You might be able to
employee something like memcached for storing queries/answers, memcached
can be setup to share memory between different servers and so on.

I'm thinking about trying to hack memcached support into pdns (either as
a native backend or using the pipe backend) rather then using flat files
or database backend as a method of updating secondary servers etc.

In theory it should get away from some of the problems we (e164.org) are
currently facing.

Database backends are "nice" from an interface perspective but once the
number of records increase greatly performance really starts to fall
off, no matter how well you tune the backend DB etc etc etc the only
cure is faster/bigger hardware.

For us DB replication isn't an option for mostly non-technical reasons,
but this still falls into the problem above.

Using the bind backend seemed like a nice option and had great speed
improvements etc over the DB option, but our zone is updated often
meaning PDNS needs to be reloaded often which as the zone grows takes
more and more time and we're back to the same issue as before.

I still need to play with memcache some more to see if there would be
any real world gains, and of course you would still need to get memcache
seeded to begin with, although this is only a problem with the authority
part of memcache, not the recursor and it should work perfectly well for
the task of a cluster of systems.

-- 

Best regards,
 Duane

http://www.freeauth.org - Enterprise Two Factor Authentication
http://www.nodedb.com - Think globally, network locally
http://www.sydneywireless.com - Telecommunications Freedom
http://e164.org - Because e164.arpa is a tax on VoIP

"In the long run the pessimist may be proved right,
    but the optimist has a better time on the trip."