Remarks and questions on SLS draft

From: Pierrick.Cheval@space.alcatel.fr
Date: Thu Oct 26 2000 - 16:07:55 CEST

  • Next message: Gerard.Gastaud@alcatel.fr: "Re: [tequila/sls] Remarks and questions on SLS draft"

    It is good to read this kind of draft when you are confronted with Access
    Services defintion.
    Good luck for the BOF. her-after are some quick remarks, in the coming days I
    will try to map the services class we are working on with what is in the draft.

    Remark/questions on Draft-tequila-diff-serv-sls-00

    General :
    Title "Service Level Specification Semantics,
    Parameters and negociation requirements"
    I like the title but either diff-serv is missing in it
    or you should narrow the goal or you should put
    for further work other technical notions which
    are quite often in actual SLAs as redundancy,
    monitoring, delay for problem notification.....

    Other services as dial-up have additional indicator as
    "successfull access/availability" or percentage of
    rejected/released call. It is probably easier to
    narrow the actual document to "always on type" of
    service (permanently physically connected).

    Detailed points:
    &3.1 Excluding the many-to-many communication is
    pretty limiting, How do you decribe a) a VPN with
    network-wide priority or reserved global capacity; b)
    wholesaling transport service sold per global capacity
    (which is may be considered as a bigger VPN).

    &3.5 Why excluding optional quantile for throughput ?
    &3.5 Qualitative : You should explain why you need to
    define qualitative performance (which does not fit
    well with guarentee isnt it ?).
    The delay and loss sub-class is not so easy to understand.
    Green,yellow,red is associated with the 3-color marker concept
    Gold, silver, bronse is usually associated with a full class
    i.e. the combination of priority,loss, delay........
    Do you think that low-delay/high-loss the same as a Service associated
    whith a low delay at the detriment of loss ( dont
    deliver data if delay exceeds an order of Nmsec) and
    is low-loss/high-delay the same as a service with a
    low loss at the detriment of delay.
    I agree it is probably an endless discussion but SLS should allow
    many different service implementations otherwise it will be rejected by
    operators.
    I think trafic priority or some relation to throughput
    is also needed as a qualitative indication.
    Finally I imagine all that qualitative discussion
    is targetted to signalling-negociation, the problem here is not
    technical but marketing, I dont imagine any operator
    writing that he is selling a high-delay/high-loss
    service.
    Why not starting simple by signalling just the
    name of the service which is required (for inter-domain as-well)
    and optionally a range of indicative values
    (e.g. target or average value).

    Pierrick

    ALCATEL SPACE INDUSTRIES
    Skybridge -- Director Access Services
    Tel : O1 4652 4027 / Fax : 01 4652 4795
    Porte : 5 rue Noël Pons 92734 Nanterre Cedex France / E-Mail :
    pierrick.cheval@space.alcatel.fr

    This message is intended only for the use of the Addressee and may contain
    information that is privileged and confidential.
     If you are not the intended recipient, you are hereby notified that any
    dissemination of this communication is strictly prohibited.
     If you have received this communication in error, please erase all copies of
    the message and notify us immediately.



    This archive was generated by hypermail 2b29 : Thu Oct 26 2000 - 16:10:21 CEST