>>>>> On Wed, 16 Oct 2002 09:49:30 +0300 (EEST), 
>>>>> Pekka Savola <[EMAIL PROTECTED]> said:

> The problems are likely to be seen with high use of UDP/ICMP, e.g. 
> sending a Gigabit/FastEthernet full of UDP and checking how much that 
> affects performance.

(if you're not requiring to add caching as MUST or SHOULD, I don't
care, but) such an environment will cause other performance issues as
well.  I guess you cannot convince others to make an additional
requirement just by warring about a special case.
 
BTW (this is almost an off-topic though): one of practical examples of
this is a heavily busy DNS server.  However, the source address
selection does not matter at least for a widely used implementation,
BIND.  It explicitly specifies the source address to make it sure that
the query's destination address equals to the reply's source address.

>> As a result, I don't think it is necessary to require (either SHOULD
>> or MUST) caching in the address selection draft.  I don't oppose to
>> mentioning the fact that caching may improve the performance, though.

> I don't think we should require anything: if an implementation wants to do 
> it the simple way, that's fine.  But I believe there should be some more 
> statements in the Implementation Considerations section, at least.

If you're not requiring anything, I'm just fine.  I made comments here
because you actually seemed to require caching in the address
selection draft.  And, it seems to me that's why others objected to
your messages (at least to some extent) in this list.

I don't see the need for the additional consideration in the
Implementation Considerations section either, but I don't oppose to
mentioning this there.  Considering implication much is in general a
good thing.  Perhaps this is up to the author (who have not appeared
in this thread...).

                                        JINMEI, Tatuya
                                        Communication Platform Lab.
                                        Corporate R&D Center, Toshiba Corp.
                                        [EMAIL PROTECTED]
--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page:                      http://playground.sun.com/ipng
FTP archive:                      ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------

Reply via email to