comparing management standards (was Comments Question about draft-xu-cops-push-00.txt)

Hi David,

(I changed the subject line since this is not about the cops-pr
proposal anymore)

(I reduced the CC's to the ops-area and the nmrg mailing lists, hoping
that will reduce duplicate mailings, I BCC'd the other addresses, to
alert them they should join the ops-area list or the nmrg list for
this discussion.) 

So far we have not done a very good job of comparing and contrasting
IETF NM standards, and providing guidelines of how they might operate
together in a system. 

I am editing an OPS area "Guidelines for Operability and
Manageability" draft that currently includes sections for describing
the applicability of IETF NM protocols and IETF NM data models.
This document may be similar to what you are describing, but to avoid
the "boiling the ocean" problem, we are focusing on comparing and
contrasting IETF standards for NM, before we consider expanding it to
include NM-related standards of other SDOs. 

As we develop the operability and manageability guidelines draft, we
may split these sections into  a second draft specifically for
comparing and contrasting available technologies. We plan to use a
bottom-up approach of understanding and documenting how IETF NM
protocols are **actually deployed** to manage real IP-based networks
today, so we can understand current practices better. It would be
helpful if other SDOs provided similar documentation of how their
NM-related standards are actually deployed to manage real IP-based
networks today. 

Such a focused approach does have its potential pitfalls, of course;
by focusing only on the IETF NM standards, in a bottom-up approach, we
might miss the bigger picture espoused by other SDOs that prefer a
top-down approach.

We are faced with a lack of quantitative research into how NM
protocols (standard or not) are used today in real networks, so it
will be hard to compare and contrast based on quantitative measures.
We will probably need to base the document we are producing on
anecdotal reports and empirical data from small research networks
until the research community can get enough cooperation from
commercial interests to be able to study patterns in commercial
enterprise and service provider networks. 

The current work being done by the NMRG and others to generate
quantitative data, and to address the concerns of commercial interests
about the sharing of data (such as anonymization), hopefully will lead
to better information in the future.

David Harrington
[email protected] 
[email protected]
[email protected]


> -----Original Message-----
> From: [email protected] 
> [mailto:[email protected]] On Behalf Of Durham, David
> Sent: Saturday, March 17, 2007 7:23 PM
> To: [email protected]; xuheyuan
> Cc: [email protected]; [email protected]; 
> Romascanu,Dan (Dan); Hexian Huang; Tom-PT Taylor; 
> [email protected]; [email protected]; Tina Tsou
> Subject: RE: [nmrg] Re: Comments Question about 
> draft-xu-cops-push-00.txt
> 
> If the goal is to use XML, then why reinvent the wheel on
webservices
> for management outside of the W3C and DMTF? 
> SNMP, COPS, DIAMETER, NETCONF... Where does it end?
> 
> Is there an RFC that does a thorough analysis of the alternatives
> (IETF's and other standards bodies' included) in terms of 
> capabilities,
> latency, bandwidth, complexity, backward compatibility and other
such
> quantitative metrics? If not, this might be a helpful activity as
the
> mailing list discussions quickly become crystal ball gazing and
> philosophical in nature. Then folks can figure out if their 
> requirements
> make a better fit for one protocol over another. 
> 
> -Dave
> 
> > -----Original Message-----
> > From: [email protected]
> [mailto:[email protected]]
> > On Behalf Of Juergen Schoenwaelder
> > Sent: Thursday, March 15, 2007 7:47 AM
> > To: xuheyuan
> > Cc: Tina Tsou; [email protected]; [email protected]; 
> Romascanu,
> Dan
> > (Dan); Hexian Huang; Tom-PT Taylor; [email protected]; ops-
> > [email protected]
> > Subject: [nmrg] Re: Comments Question about 
> draft-xu-cops-push-00.txt
> > 
> > On Thu, Mar 15, 2007 at 10:19:18PM +0800, xuheyuan wrote:
> > 
> > > At that time the Re reference point, merged into Rw 
> reference point
> > > later, controlled the IP edge, actually an IP-IP gateway. In the
> > > current network, most existing IP-IP gateways are routers, which
> > > support COPS already. In reality, using COPS is the 
> quickest way to
> > > do that as described in approach 2 in this draft.
> > 
> > I have routers that support NETCONF in my lab and no routers
> > supporting COPS-PR or COPS.  Perhaps this is why our views of the
> > world are somewhat different.
> > 
> > > COPS-PR by nature does policy control, Rw interface mainly does
> > > policy control dynamically, not just push configuration
> > > information. Besides, in 2005,NETCONF was a pretty new 
> protocol, and
> > > few existing routers supported NETCONF.
> > 
> > I never understood the distinction between configuration and
policy
> > information when it comes to the device interface. The PDP is the
> > policy engine and what comes out of this engine as provisioning
> > information is for me not really distinct from configuration data
> > coming out of other systems (e.g. customer provisioning systems).
> > 
> > I agree that NETCONF was pretty new in 2005 and it still is in
2007.
> > COPS-PR is older but as far as I can tell not too widely
implemented
> > and the number of people within the IETF who do understand COPS-PR
> > well is like to be somewhat smaller than the number of people
within
> > the IETF who understand NETCONF and my impression was that COPS-PR
> > awareness declines.
> > 
> > > As RACF supports both fix and mobile, push and pull modes, in
the
> > > real networking, PD-FE may possibly support both modes, 
> and in pull
> > > modes, COPS is very effective, so, in push mode, if 
> approach 2 COPS
> > > could be adopted, it will simplify the implementation.
> > 
> > The implementation simplification is a relative thing (if you have
> > COPS, using COPS is simple, if you have NETCONF, using NETCONF is
> > simple). Not sure how to make a decision on this.
> > 
> > Again, I like to see the discussion raised to the technical level
to
> > first understand if NETCONF could do what you need it to do
equally
> > well or whether there is any technical feature that really
> > distinguishes COPS-PR from NETCONF.
> > 
> > /js
> > 
> > --
> > Juergen Schoenwaelder		 Jacobs University Bremen
> > <http://www.eecs.iu-bremen.de/>	 P.O. Box 750 561, 28725
Bremen,
> > Germany
> > --
> > !! This message is brought to you via the `nmrg' mailing list.
> > !! Please do not reply to this message to unsubscribe. To 
> unsubscribe
> or
> > adjust
> > !! your settings, send a mail message to
> <[email protected]>
> > !! or look at https://mail.ibr.cs.tu-bs.de/mailman/listinfo/nmrg.
> 
> -- 
> !! This message is brought to you via the `nmrg' mailing list.
> !! Please do not reply to this message to unsubscribe. To 
> unsubscribe or adjust
> !! your settings, send a mail message to 
> <[email protected]>
> !! or look at https://mail.ibr.cs.tu-bs.de/mailman/listinfo/nmrg.
> 



_______________________________________________
OPS-NM mailing list
[email protected]
https://www1.ietf.org/mailman/listinfo/ops-nm