Re: High availability/contingency for DNS?

Very much second what Matt said there. DNS is inherently scalable.

I have had some experience of running the authoritative name servers for .uk: BIND is highly scalable and DNS is inherently redundant.

strictly speaking incorrect zone data is not necessarily the same as a negative response resulting in NXDOMAIN.

But caching resolvers cache positive responses (as in this domain is in the zone file - whether what it points to is what you want or not) as well as negative (this domain is not in the zone file).

There are separate ttl's determining how long this information is cached for, and by default it is on the order of days.

jason.

--
http://jarneil.wordpress.com

On 15 Nov 2011, at 21:22, Martin Bach wrote:



> -------- Original Message --------

> Subject: 	RE: High availability/contingency for DNS?

> Date: 	Tue, 15 Nov 2011 15:16:39 -0600

> From: 	Zito, Matthew <Matt_Zito_at_bmc.com>

> To: 	development_at_the-playground.de <development_at_the-playground.de>

> 

> 

> Maybe you can forward this to the list for me, as my list setup is borked at the moment.

> 

> In any case, DNS is built to be redundant fundamentally.  First off, the local resolver on a machine can be configured to look at multiple servers - if one is down, it fails over to the next one automatically.  Then, the authoritative servers for zones can be configured with multiple sets - if one returns with an error, servers and resolvers will roll on to the next authoritative machine.

> 

> The only place you run into real errors is when a machine believes it is authoritative for a zone but has incorrect data.  This can get quite messy, as caching servers will start to hang on to the negative responses, making it difficult to troubleshoot and fix.  This isn't the protocol's fault, though, it's really the implementation behaving badly.

> 

> Matt

> 

> 

> 

> --

> http://www.freelists.org/webpage/oracle-l

> 

> 


--
http://www.freelists.org/webpage/oracle-l