RE: [Simple] I-DAction:draft-ietf-simple-interdomain-scaling-analysis-03.txt

["Joe Hildebrand" <JHildebrand at jabber.com>]

> Did you look at the numbers in section at section 2.10 at the 
> draft? Note that the claimed comparison was made for 3K 
> message while using partial notifies. 

Yes.  I'm saying that XMPP effectively always uses partial notifies, so we won't have 3k messages, even with 3k of presence data.

> Maybe I should not have said "e.g. XMPP" since in XMPP you 
> can claim that you will never use such a big presence 
> document due to the use of getting other parts of the 
> presence document via other ways like PEP. 

If you have another presence protocol in mind, like IMPS/WV, I don't have any objection to changing the e.g.

> However, it seems to me that at the end of a day data is 
> data, it does not really matter if you are going to get it 
> from the server of via other ways. 

It's a question of how often you get which subset of the data.  If you get a small subset of the data, only when it changes, rather than a full document for any change, it has a large impact on bandwidth.  This is the same reason that partial-notify exists.  

Even with the caveat that in XMPP you have to transmit the entire PEP item (a collection of related data, rather than just the changed element), the overhead of the SIP headers will almost always be greater than the item overhead, since PEP items tend to be very small.

-- 
Joe Hildebrand


_______________________________________________
Simple mailing list
Simple at ietf.org
https://www1.ietf.org/mailman/listinfo/simple