RE: netconf WG charter proposal

["Harrington, David" <dbh@enterasys.com>]

Hi Andy,

I'm sorry my question upset you. I was just asking a question I hadn't seen discussed.

To answer your questions,

The operators apparently believe the CLI is the best current mechanism for configuration management. It is the interface they use the most. It is best for the reasons you cite - it allows full, initial, and OOB configuration, is typically the only interface available to do that, and it is widely available on vendors' products (albeit not standardized). 

I didn't ignore the distinction between a human and programmatic interface. The fact is, the CLI may be designed to be a human-interaction interface, but it has also been harnessed to be a programmatic interface using scripting. For the same reasons already cited, it appears to be the best current interface for programmatic configuration as well, even though it is not standardized.

Getting vendors to agree would be no different for a standardized CLI as it was for standardized SNMP or will be for standardized XML. CLI is just another interface that would benefit from standardization. The problem to be solved is to convince vendors to agree to a standard for configuration - any standard! The main reason vendors have not agreed to a standard interface is they want to develop unique products, and unique products often require unique configuration commands/parameters/data models. Using XML isn't likely to change that basic economic point.

I suspect from your tone that you think I oppose XMLConf. I don't. I don't oppose the use of XML at all; in fact I think XML has great promise as a language for integrating device operational databases (config, state, and statistics) with databases, directories, and for sharing data between applications. I strongly believe it is a good direction to be heading and have so advised my company.

I am very aware of the w3c and all the work that is being done for XML. I recognize that XML has a ton of features that could be used to solve the problems we are trying to address. I see people touting the benefits we will get from data modeling with XML Schema, and validation using XSLT, and separation of meta-data and data, and on and on... That is my concern. 

Occam's Razor admonishes us to choose from a set of otherwise equivalent models of a given phenomenon the simplest one.

As a standards developer I have concerns that using XML and all the assorted baggage that comes with it may be overkill. I have seen overkill many times in the IETF, and I think people should always ask "could this be done in a simpler way?". 

As a vendor representative, I have concerns about how much stuff will need to be supported in our devices. My company has an interest in making sure somebody asks the question about whether there is a simpler way.

Nobody seems to have asked the question. 
 
dbh

-----Original Message-----
From: Andy Bierman [mailto:abierman@cisco.com]
Sent: Wednesday, April 09, 2003 5:44 PM
To: Harrington, David
Cc: xmlconf@ops.ietf.org
Subject: RE: netconf WG charter proposal


At 03:25 PM 4/9/2003 -0400, Harrington, David wrote: 
>Hi, 
> 
>I haven't seen anybody actually raise this point. 
> 
>If the CLI is currently the best mechanism to do configuration, but suffers from lack of standardization, why are we moving to XML? Why aren't we just addressing the problem of standardization of the CLI?
Who said CLI is the best interface for configuration management? 
On many devices, it is the only interface that allows full, 
initial, and OOB configuration.  Also, you are ignoring the 
distinction between a human interface and a programmatic interface. 


>Wouldn't it be simpler to standardize a CLI syntax and data model using existing ascii-style text and the associated tools, than to standardize an XML syntax and data model from scratch?
Huh?  It won't be simple at all to get vendors to agree 
on the One True CLI Design.  We aren't starting from scratch 
with XML. Look at http://www.w3c.org/ at the massive amount 
of work already done on XML standards. 
You want to write an SMI for CLI syntax?  It is so much 
easier to use a standard that meets our needs and is already 
deployed.  We need the data modelling capabilities of XML Schema, 
the UTF-8 encoding for internationalization requirements, the 
validation features of XSDs, the separation of meta-data and data, 
and on and on... 


>dbh 
Andy 


>>Andy>    - Uses a data representation that is easily manipulated 
>>Andy>      using non-specialized text manipulation tools 
>> 
>>So XML is out if I take that text serious. XML documents have 
>>structure and to really operate on them, you need to have tools that 
>>understand the structure. The good news is that there is an increasing 
>>amount of such tools. 
> 
>>I will change the text to say XML will be used 
> 
>-- 
>to unsubscribe send a message to xmlconf-request@ops.ietf.org with 
>the word 'unsubscribe' in a single line as the message text body. 
>archive: <http://ops.ietf.org/lists/xmlconf/> 

--
to unsubscribe send a message to xmlconf-request@ops.ietf.org with
the word 'unsubscribe' in a single line as the message text body.
archive: <http://ops.ietf.org/lists/xmlconf/>