Re: gethostbyname (was Re: (DNS) draft wording for W.G. review. ) from Anawat Chankhunthod on 1996-03-28 (ietf-http-wg@w3.org from January to March 1996)

From: Anawat Chankhunthod <chankhun@catarina.usc.edu>
Date: Wed, 27 Mar 1996 19:38:08 -0800
To: Paul Leach <paulle@microsoft.com>
Cc: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
Message-Id: <199603280339.TAA09335@catarina.usc.edu>

> If the API change will require new implementations of gethostbyname(),
> why not get it (and friends) to do the caching, or reasonably efficient
> access to the local resolver's cache, instead of pushing this into the
> app layer? 

Agree. We need client side (resolver's client) caching. I think we can't afford
to make a network conversation, even on udp on every single dns lookups. 

Also, if resolver library does a caching/nb-lookup, we can feel more 
comfortable about DNS security. As least we will be using the same algorithm,
not 100 different caching scheme by 100 different implementations/programmers.

Anawat

Received on Wednesday, 27 March 1996 20:45:37 UTC