[Ietf-not43] oss implementation release

[Rick Wesson]

On Tue, 29 Oct 2002, Andrew Newton wrote:

> Yoah!  Is this a context switch to the requirements?
> 
> Query distribution is just giving the client a list of places to go 
> search.  As for processing requirements, they shouldn't be any greather 
> than they are today.  In fact, by structuring the queries, processing 
> may be reduced by limiting the number of indexes that must be consulted.

my question is if the query for a contact must be distributed to all
servers then some servers with few registrations/contacts will be doing
alot of processing to respond with a negative result.

I feel that queries for contacts will be a burdon on the servers with few
contacts. When we have requirements that require all participiants to
provision equal bandwith and cpu for processing even when they know they
will rarely have a positive result to return.

We need to discuss the requirement on support of searches for objects that
require the client to query every participating server where no
authortaive index exists.

-rick