Re: draft-ietf-tewg-measure-05.txt comments

["Dr. Richard W. Tibbs" <tewg@oakcitysolutions.com>]

Jim, Jerry, etal:
Your comments are indeed helpful.  I have agreed with Wai to try a new 
version of our draft, based on the feedback we have had in the last 
couple of weeks.  It will take me  a while to do what is proposed -- 
which is significant "thinning" and re-organizing of the document --  in 
your comments, but I am willing.    As I do this, I may need to clarify 
some of your comments and I will mail the list to do so.  
Yours,
Dr. Richard W. Tibbs



Ash, Gerald R (Jerry), ALABS wrote:

> Jim,
>
> Your detailed comments and suggestions are very helpful, and should help improve the quality of the document.  
> I have some comments below on Tom's and your comments.
>
> Thanks,
> Jerry
>
> Jim> Flow measurement is a deep-hole, we should either defer to other
> Jim> standardization efforts on this, or we need to be a lot more
> Jim> concrete on what we need to know here.	
>
> Tom> 	Defer to the IPFX WG that is tackling this problem right now.
>
> Let's be more concrete in the TEM draft re flow measurement requirements.  I don't see a lot on TE measurement requirements in the IPFIX charter or requirements document http://ietf.org/internet-drafts/draft-ietf-ipfix-reqs-09.txt.  Let's be sure all the requirements get stated.
>
> Jim> " Need for higher order statistics... "  push it to the offline hosts
> Jim> "Need for packet-sampled..."  "Need for offline bulk file 
> Jim> transfer..." The need needs to be better justified or removed.
>
> Tom> 	I asked to have the offline stuff added. For the
> Tom> collection of a traffic matrix, as you know, this 
> Tom> entails the collection of a lot of data especially on
> Tom> big core routers. It is preferable to use offline
> Tom> bulk file transfer mechanisms such as NetFlow (being
> Tom> standardized in IPFX now) or offline SNMP file transfers
> Tom> of statistics snapshots.  This reduces the overhead of
> Tom> the box having to service network management requests,
> Tom> and also reduces the amount of network mgmt overhead
> Tom> required to fetch the same amount of data if you 
> Tom> collect your data as one big batch at some longer periodic
> Tom> interval.
>
> Quite unclear that moving massive amounts of data for offline processing is always 'preferable' to online statistics to meet all requirements.  Data collection/summarization in the 'big core routers', e.g., on a node-pair basis, is very desirable for traffic matrix derivation in some applications.