[Pdns-users] Re: PowerDNS performance compared to other nameservers

[Mike Benoit]

On Tue, 2003-12-02 at 23:58, Stephane Bortzmeyer wrote:
> On Tue, Dec 02, 2003 at 07:45:02PM -0800,
>  Mike Benoit <ipso at snappymail.ca> wrote 
>  a message of 78 lines which said:
> 
> > Since that is what Bind and most other name servers do, read _all_
> > the zones in to memory (cache) on startup, before they answer any
> > queries.  Add 100,000 zones to Bind, and time the startup process +
> > 1,000,000 queries and I bet PowerDNS would pull ahead! This is the
> > main reason we switched to PowerDNS,
> ...
> > You simply can not do this to any scale (ie: >50,000 zones) with
> > Bind.
> 
> Since it seems that you know only BIND and PowerDNS, may I draw your
> attention to the actual comparison, where the clear winner,
> performance-wise, was nsd?

I wasn't disputing the winner as much as I was disputing the entire
comparison as "unfair" and incomplete. Just because NSD comes out way on
top doesn't mean you shouldn't bother to minimally optimize BIND and
PowerDNS to make the comparison fair. Thats as bad as Microsoft paying
someone to run a benchmark where their software always happens to come
out on top. Why bother spending time to optimize competing products when
you know ahead of time which program is going to "win". Benchmarks like
this are next to useless.

I've never used NSD, but I'm gonna guess it doesn't have half the
features that BIND does, and I know both of them don't offer a lot of
the things that PowerDNS does. So obviously the more features you add,
the less performance your going to get. Just the nature of PowerDNS,
with its generic backend design probably makes it so it will never beat
BIND in raw performance. But that doesn't matter to most people who are
considering PowerDNS for their name servers, because they are looking
for flexibility that neither BIND or NSD offer, not raw speed. 

How many people truly need 28,000q/s, or even 9000q/s?  


>  
-- 
Mike Benoit <ipso at snappymail.ca>