Re: [tequila/sls] comments on SLS scope

From: Abdallah Rayhan (arayhan@nortelnetworks.com)
Date: Fri Mar 30 2001 - 22:19:53 CEST

  • Next message: Panos Georgatsos: "Re: [tequila/sls] comments on SLS scope"

    I believe the purpose of what we doing here is to come up with
    something that is efficient and practical to implement. CIM might
    be favored from a standard-like architecture but when it comes
    it to implementation, from my experience, it is not the first
    choice of designers. Furthermore, it does not represent a data
    model that is particularly compatible with any platform, thereby
    making implementation difficult and non-deterministic. The CIM
    schema is also less optimal for representation of data for
    management purposes. INHO, CIM's use of XML is sub-optimal for
    a protocol purposes. I think an infomodel/XML Schema designed
    with this in mind would be too heavyweight for protocol purposes.

    As for storage in regards to a comment made by Marcus, addressing
    the storage issues is important and at some point of time we need
    to address that, although not necessarily in this WG. Thinking
    about the "conceptual" model of the protocol (e.g. entity a talks
    to entity b with the following parameters) makes sense. In fact I
    usually need to define protocols to help translate an information
    model from one datastore to another, whereas if I start with the
    protocol then each storage model will be forced to support the
    same protocol model.

    The question is: Should we define a representation of an SLA,
    then figure out how to transport it? Or should we define the
    protocol, and then optimize the storage? I think the latter
    makes sense in protocol design. Take the policy LDAP schema for
    instance... it is not optimized storage for the delivery protocols
    (SMTP, COPS-PR). It is however optimized for LDAP representation,
    which presumably makes it convenient for LDAP-speaking devices
    and management clients.

    my 2 cents

    regards
    Abdallah

    LAVERGNE Laurent DvSI/SIReS/LAN wrote:
    >
    > I don't understand the point. Is SLS looking for an modeling language to
    > describe the SLS information (for real time negociation) or is it looking
    > for an existing information model to plug to?
    >
    > I think I agree with the first part of the previous message and don't
    > understand the need of CIM (or even PCIM), someone could perhaps explain me
    > this.
    >
    > regards
    > L.Lavergne, France Telecom



    This archive was generated by hypermail 2b29 : Fri Mar 30 2001 - 20:45:12 CEST