[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: Netconf Event Message as Working Group Document
in-line
Hector
-----Original Message-----
From: David B Harrington [mailto:dbharrington@comcast.net]
Sent: Tuesday, November 08, 2005 2:13 PM
To: 'Andy Bierman'; Hector Trevino (htrevino)
Cc: 'Wijnen, Bert (Bert)'; j.schoenwaelder@iu-bremen.de; 'Sharon
Chisholm'; netconf@ops.ietf.org
Subject: RE: Netconf Event Message as Working Group Document
Hi,
I have an opinion.
I think there is real value in being able to correlate snmp
notifications with syslog messages, and likely with other event-driven
communications.
HT: Agree. One of the appendices mentions this and has a description of
an Event Id field for this purpose.
I have concerns with developing another type of notificaiton before
integrating with existing solutions. The ISMS WG was chartered because
operators expressed an interest in not having to deal with yet-another
solution for security. I have not seen strong demand for yet-another
solution for notifications.
HT:
The event draft invents a new format with new content. There appears to
be some correlation with ITU-style of events management, and an effort
to reuse designs from ITU and other SDOs may be an important benefit,
but I'd rather see us address reuse of/correlation with our own
standards first.
HT: I'd say that both re-use of non-IETF & IETF standards was attempted
in the doc.
My professional role is typically as an NM architect for an equipment
vendor, and I would have trouble evangelizing the adoption of this
technology in my employer's devices and application without having a
much stronger case made for the benefits of this new work compared to
existing notification solutions.
HT: OK
If a proposed event messaging solution provided a means for carrying
SNMP notifications and syslog messages and a means to correlate them
(e.g. with timestamps or event IDs), that would be a benefit I could try
to sell as a benefit to customers. A tie-in to ITU-style events and
alarms might be an additional benefit I might be able to evangelize. I
am not convinced the current draft provide thos benefits, but it is
possible they could be added if consensus was to do so.
HT: The doc does not address SNMP but does address syslog (appendix
proposal). The tie to ITU event definitions is also there. Perhaps
things could be clarified/expanded upon but the subjects are covered.
One of the justification for netconf is its textual format - explicitly
because operators want to be able to save configurations in source code
control systems that cannot handle binary, so those configurations can
be distributed to multiple endpoints after modification using standard
text-editing tools. I don't see how defining notifications/events in XML
helps solve that problem.
HT: NETCONF does not specify the payload of the operations - it could be
XML or it could be plain text. The operations are in XML and assuming
that over time a data/info model is defined using XML then it makes
sense to also define notifications in XML. The important points are that
a) everything is defined in the same manner and b) that it is well
defined/structured (schema) and simplifies handling/parsing of the
message.
I don't want to say we should not accept these documents, but neither do
I want to say we should. I was one of the attendees to the editing
session, and I went to understand what progress has been made, as did
others. I don't think there has been adequate demonstration of the need
for this work; more work is needed. I have difficulty asking a product
manager to add this new feature to equipment without some proof of value
or demand, especially since netconf itself has not yet been proven
useful to or in demand by operators.
HT: OK
So far, netconf has not been successfully tested for interoperability,
largely because there are no vendor-neutral data models to test with.
I think that is more important to establish some basic vendor-neutral
data models than to add new features. The success of SNMP was partly
because there were vendor-neutral mib modules to work with.
HT: Mostly agree. The models are necessary if any sort of
interoperability is ever to be achieved. The notifications are needed to
provide a complete solution and I think both are needed.
But I have a vendor's perspective, and we really should get the operator
view. Do operators want another information/data model for events? Do
they want the correlation between SNMP, Netconf, and syslog events? Is
this more important than establishing a mechanism for standardizing data
modeling? Is it important to operators to be able to access SNMP data
via netconf?
HT: Based on what I've seen, people use syslog because there is nothing
better that has as complete coverage. Where available, people want to
use SNMP because things are well defined/structured (data model part).
So, something is needed that covers both. Yes, I know that defining the
events alone won't solve the problem but I think it's a step in the
right direction.
Hector
David Harrington
dbharrington@comcast.net
> -----Original Message-----
> From: owner-netconf@ops.ietf.org
> [mailto:owner-netconf@ops.ietf.org] On Behalf Of Andy Bierman
> Sent: Tuesday, November 08, 2005 11:11 AM
> To: Hector Trevino (htrevino)
> Cc: Wijnen, Bert (Bert); j.schoenwaelder@iu-bremen.de; Sharon
> Chisholm; netconf@ops.ietf.org
> Subject: Re: Netconf Event Message as Working Group Document
>
> Hector Trevino (htrevino) wrote:
>
> >Yes, you may assume that.
> >
> Is this syslog, SNMP notification, or something new?
>
> IMO, notifications in the original charter are not clearly enough
> specified to be useful in this debate. The first debate I want to
> have is whether NETCONF should invent a new event system or
> adopt/invent NETCONF XML encodings of syslog and/or SNMP
> notifications.
>
> Anyone one the list with an opinion, please speak up!
>
> >
> >Hector
> >
> >
>
> Andy
>
> >
> >
> >-----Original Message-----
> >From: Wijnen, Bert (Bert) [mailto:bwijnen@lucent.com]
> >Sent: Tuesday, November 08, 2005 11:02 AM
> >To: Hector Trevino (htrevino); j.schoenwaelder@iu-bremen.de
> >Cc: Sharon Chisholm; netconf@ops.ietf.org
> >Subject: RE: Netconf Event Message as Working Group Document
> >
> >So may I assume you guys are implementing on top of an existing
Cisco
> >NetConf implementation?
> >
> >Bert
> >
> >
> >
> >>-----Original Message-----
> >>From: owner-netconf@ops.ietf.org
> [mailto:owner-netconf@ops.ietf.org]On
> >>Behalf Of Hector Trevino (htrevino)
> >>Sent: Tuesday, November 08, 2005 18:44
> >>To: j.schoenwaelder@iu-bremen.de
> >>Cc: Sharon Chisholm; netconf@ops.ietf.org
> >>Subject: RE: Netconf Event Message as Working Group Document
> >>
> >>
> >>
> >>Hi, that would be various people at Cisco who are interested in
> >>NETCONF and think notification support is needed.
> >>Hector
> >>
> >>
> >>-----Original Message-----
> >>From: Juergen Schoenwaelder [mailto:j.schoenwaelder@iu-bremen.de]
> >>Sent: Tuesday, November 08, 2005 12:57 AM
> >>To: Hector Trevino (htrevino)
> >>Cc: Sharon Chisholm; netconf@ops.ietf.org
> >>Subject: Re: Netconf Event Message as Working Group Document
> >>
> >>On Mon, Nov 07, 2005 at 07:43:16PM -0800, Hector Trevino
(htrevino)
> >>wrote:
> >>
> >>
> >>>Sharon,
> >>>
> >>>Agree with your suggestion. We'd like to see this work move
> >>>
> >>>
> >>ahead in
> >>
> >>
> >>>the NETCONF WG.
> >>>
> >>>
> >>Who is "we" here?
> >>
> >>/js
> >>
> >>--
> >>Juergen Schoenwaelder International
> University Bremen
> >><http://www.eecs.iu-bremen.de/> P.O. Box 750 561,
> >>28725 Bremen,
> >>Germany
> >>
> >>--
> >>to unsubscribe send a message to
> netconf-request@ops.ietf.org with the
> >>
> >>
> >
> >
> >
> >>word 'unsubscribe' in a single line as the message text body.
> >>archive: <http://ops.ietf.org/lists/netconf/>
> >>
> >>
> >>
> >
> >--
> >to unsubscribe send a message to netconf-request@ops.ietf.org with
> >the word 'unsubscribe' in a single line as the message text body.
> >archive: <http://ops.ietf.org/lists/netconf/>
> >
> >
> >
> >
>
>
> --
> to unsubscribe send a message to netconf-request@ops.ietf.org with the
> word 'unsubscribe' in a single line as the message text body.
> archive: <http://ops.ietf.org/lists/netconf/>
>
--
to unsubscribe send a message to netconf-request@ops.ietf.org with
the word 'unsubscribe' in a single line as the message text body.
archive: <http://ops.ietf.org/lists/netconf/>