Re: draft-zeilenga-ldup-harmful-00.txt

At 09:47 AM 5/6/2003, Mark C Smith wrote:
>Kurt D. Zeilenga wrote:
>
>>At 07:32 AM 5/6/2003, John McGarvey wrote:
>>
>>>IMHO, this draft takes a very extreme position, stated here:
>>
>>This draft only reiterates the position which RFC 2251 and
>>X.501(93) took, and which LDAPBIS is presently taking.  LDUP
>>takes an extreme position that an extension to LDAP can alter
>>the LDAP service model in ways which effect implementations
>>not supporting the extension.  Such proposals should not be
>>treated as extensions, but as NEW protocols.
>
>I disagree. And like John, I also disagree with this text from draft-zeilenga-ldup-harmful-00.txt:
>
>  The X.500/LDAP information and service models does not support
>  multi-master replication and cannot be altered to support multi-master
>  replication without causing great harm.

CI assume you are disagreeing with the latter portion of sentence
as clearly the X.500/LDAP information and service models do not
support multi-master replication.

John suggests changing the service model.  I argue that this will
cause great harm.  It certainly will break any application which
relies on this aspect of the LDAP/X.500 service model.

>I do think it would be worthwhile for the LDUP documents to address the scenarios you spell out in your -harmful draft. John suggested some possible solutions, all of which have likely been used in one or more of the existing, widely deployed LDAP implementations that happen to support multimaster replication (MMR).

Again, John suggests changing the LDAP service model.  That, I believe,
is beyond the scope of an LDAP extension and can only be done through
revision of the LDAP "core" technical specification.

>The fact is, there is a lot of demand from customers/deployers for MMR.

There is lots of customer demand to maintain the current
LDAP service model.  There is also significant demand to
expand that model to support LDAP Transactions.

While I generally support extending LDAP to support new
demands, I do not support doing so at the expense of those
who rely on LDAP being implemented as currently specified.

>If the IETF does not define anything in this area, then the directory community will continue to have no common specification for MMR. 

The IETF is not likely to define anything in this area that is
workable.

>And that will likely cause great harm to LDAP's future.

LDAP is what it is.  If the demands are not met by the current
specification and cannot be met through appropriate extensions,
then it's time to design a new protocol (LDAPv4?) with a
service model that meets those demands.

Kurt