[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: Perspective: XML's ticking time bomb

HI John,

On the below, I must congratulate Intelliden if the company has truely
created "a model of IOS and CatOS (and a couple more OSs from cisco)".
The successful creation of a model has the potential to provide
much economic benefit to Intelliden. And if Intelliden has not quite
accomplished the achievement, they still deserve congratulations.

Now, the big issue with modelling is being able to create models
that are both workable for current products and for future products
and usages. This is hard, really hard. And in my opinion, most of
the appeal of the new management protocols that have been offered
over the last 10+ years is the reworking of the model. It is much
easier for a competent engineer (or most likely, a team) to create
an "easy to understand" and straight forward model for existing
technology that is well understood in how it works, and how it
is used, than to create models for new (and future) technology
that is not well understood, and whose usage has no operational
experience data.

Or put another way, Question: "Why is creating management
for new technologies so difficult?"  Answer: "Because you have
to correctly predict the future and make engineering choices
based on your predictions"

Now, how do I invest in Intelliden?


At 07:18 PM 1/6/2003 -0700, John Strassner wrote:

Hi Eliot,

I responded because I disagree with the characterization that CIM is here and can solve everything. And even though XMLCONF hasn't debated model specifics - maybe it's time it should. Even if you don't use a model, such as CIM or SID or DEN-ng, you still must organize the information in some way. Which is what Bert was alluding to (or at least, I tool it that way) - if all you have is a forest of classes, so what? That will not help solve the provisioning problem.

And a common language to do what Eliot? Are you saying that we don't need a model? Before you answer, my company built a model of IOS and CatOS (and a couple more OSs from Cisco ;-( ), and then built XSDs from those. Our product uses BOTH to provision a device.

regards,
John
 
John Strassner
Chief Strategy Officer
Intelliden Corporation
90 South Cascade Avenue
Colorado Springs, CO  80903  USA
phone: +1.719.785.0648
  FAX: +1.719.785.0644
email: john.strassner@intelliden.com
 

-----Original Message-----
From: Eliot Lear [mailto:lear@cisco.com]
Sent: Monday, January 06, 2003 6:26 PM
To: John Strassner
Cc: Andrea Westerinen; Faye Ly; Wijnen, Bert (Bert); Xmlconf (E-mail)
Subject: Re: Perspective: XML's ticking time bomb

John/Andrea,

Without entering into the MOF debate, I think you would agree with
Andrea's major point that it would be nice if we had a common language.
within the XML that could be spoken, and not just widgets, and that if
you're going to use widgets, then being able to express CIM (or pick
your favorate common information model (lowercase)) in those widgets is
a good thing.

This group (XMLCONF) has not debated model specifics -- at all.

Eliot


John Strassner wrote:
> I don't see how CIM accomplishes this. There are no managed objects
> for representing fundamental constructs, such as a physical port of a
> device or the logical interface (or sub-interface) of a device. You
> claim that a protocol endpoint is an interface - but it isn't. It is a
> general purpose communications interface. From the Network 2.6 MOF:
>
> // ==================================================================
> //     ProtocolEndpoint
> // ==================================================================
>         [Description (
>          "A communication point from which data may be sent or "
>          "received. ProtocolEndpoints link router interfaces and "
>          "switch ports to LogicalNetworks.") ]
>
> Furthermore, there is no public CIM class to represent a device
> interface. So please tell me how CIM enables the developer to get from
> an XML representation of (for example) a change in the configuration
> file of a device to a model representing that change.
>
> Finally, DTDs are useless. And WSDL is independent of XML, so I'm not
> sure why you even brought that up.
>
> regards,
> John
> 
> John Strassner
> Chief Strategy Officer
> Intelliden Corporation
> 90 South Cascade Avenue
> Colorado Springs, CO  80903  USA
> phone: +1.719.785.0648
>   FAX: +1.719.785.0644
> email: john.strassner@intelliden.com
> 
>
>
> -----Original Message-----
> From: Andrea Westerinen [mailto:andreaw@cisco.com]
> Sent: Sunday, January 05, 2003 2:30 PM
> To: Faye Ly; Wijnen, Bert (Bert); Xmlconf (E-mail)
> Subject: RE: Perspective: XML's ticking time bomb
>
>
> This is exactly what the DMTF is trying to do with CIM - create an OO
> model for managing end to end (provisioning, monitoring, faults, ...). 
> The model is tied together across management domains, vendors, and
> products. Whether the model is encoded in XML or something else is
> secondary.  The fact that concepts like systems, services, interfaces
> (protocol endpoints), etc. can be modeled and generically understood
> (have inherited properties and
> behaviors) is much more valuable than the encoding.  After all, "XML is just
> a syntax in search of a semantic."
>
> However, if you are looking for XML, then there is an XML DTD for CIM
> - and also work-in-progress regarding a CIM XML-Schema, and
> discussions of CIM-WSDL.
>
> Andrea
>
> -----Original Message-----
> From: owner-xmlconf@ops.ietf.org [mailto:owner-xmlconf@ops.ietf.org]On
> Behalf Of Faye Ly
> Sent: Sunday, January 05, 2003 8:51 AM
> To: Wijnen, Bert (Bert); Xmlconf (E-mail)
> Subject: RE: Perspective: XML's ticking time bomb
>
>
> Bert,
>
> That is a very good article.  I admit I went back to this mailing
> list's archive and got lost in the multiple mail threads.  So what is
> the conclusion on moving forward for this group?
>
> I think I tend to agree that XML is a superior language over MIB but
> the fact that we are missing 'management object' on many things such
> as -
>
> Service provisioning/ subscriber provisioning
> fault isolation that is transparent to the underlying transport method
> ...
>
> Sort of similar to the effort of snmpconf (for provisioning only) that
> is currently missing.  I actually think it is in-relevant if we do it
> using XML or the good old MIB.  The important thing is to come up with
> consensus on the management model.  If XML can help with the majority
> of the people to better understand and thus expedite the process, then
> let's go with XML.  I think this is actually the time to organize the
> effort around coming up with standards for:
>
> 1. provisioning
> 2. fault isolation
> 3. performance monitoring
> 4. othrs such as file management, upgrade and etc ...
>
> And let each group come up with the management model first, XML and/or
> MIB later?
>
> What do you think?
>
> -faye
>
> -----Original Message-----
> From: Wijnen, Bert (Bert) [mailto:bwijnen@lucent.com]
> Sent: Sunday, January 05, 2003 3:51 AM
> To: Xmlconf (E-mail)
> Subject: Perspective: XML's ticking time bomb
>
> Here is another one to take into account:
>
> Perspective: XML's ticking time bomb
>
>   http://news.com.com/2010-1071-961117.html
>
> It is a few months old... not sure how I all of a sudden
> ran into it. Oh well...
>
> Bert
>
> --
Regards,
/david t. perkins