TM Forum Community

Expand all | Collapse all

YANG like model for physical equipment onboarding

  • 1.  YANG like model for physical equipment onboarding

    TM Forum Member
    Posted Feb 02, 2021 19:15
    Hi,

    Looking for an example of a YANG model for physical equipment as described in this catalyst: https://inform.tmforum.org/catalyst/2018/05/catalyst-aims-simplify-onboarding-physical-network-gear/

    If an example can be shared that help, as I am trying to understand what this would look like.

    Thanks.
    #OpenDigitalArchitecture

    ------------------------------
    Saravana Chockalingam
    Telstra Corporation
    ------------------------------


  • 2.  RE: YANG like model for physical equipment onboarding

    Posted Feb 03, 2021 11:35
    Saravana, thank you for your interest. Some years ago I noted that tools being proposed to describe the hierarchy of software capabilities and requirements could equally well be used to describe the hierarchy of equipment in things like modular routers. Eg. A port is provided on a card but a card needs a slot, a slot is provided by a  chassis, a chassis needs a control card etc.
    Typically this information is supplied in spec sheets but if it were machine readable then there is scope for automatically making recommending the optimal collection of equipment to use when increasing the port capacity at a particular location. No doubt there other advantages as well.
    At the time I tried to specify a syntax for this based on YANG as that language was already being introduced to describe the software capabilities of the equipment. Also I choose to use IETF YANG definitions  which turned out not to be the preferred flavour.
    BT made approaches to three of the major router vendors, there needs to be something in it for them too - I would have thought build validation would be easier, but the reaction was only lukewarm. Of course, there is still the option of a CSP transcribing datasheets into machine readable format for their own internal use.

    The YANG definition I created at the time is at  GitHub - pmjordan/wayang: Proposed format for machine readable data normally found in data sheets of hardware
    but I'm afraid I no longer have the example router description which conformed to that definition.

    I think TOSCA could be used in the same way should that syntax prove to be more acceptable and I do have an example of that. Please contact me directly if you would like to see that or discuss further.





    ------------------------------
    Paul Jordan
    BT Group plc
    ------------------------------



  • 3.  RE: YANG like model for physical equipment onboarding

    TM Forum Member
    Posted Feb 04, 2021 05:35
    I used to work in the Advanced Software Engineering group in Nortel Networks during the early '90s. We applied different software techniques to different problems to see which were most effective and then turned them into products for the company.  We solved these problems over 25 years ago.

    We applied the Constraint Based Handling software technique to equipment configuration and created an application that Nortel used this for all its hardware products. We would produce a near-optimal configuration with seconds rather than an engineer taking hours and Nortel used it throughout its product lines by its product and field engineers.

    We took this a step(s) further and created a resource-based configuration language that would take a given product and produce the required network configuration to support that product; and vice-versa - given a network of equipment, what services could be supported. This allowed us to then cope with network failures and automatically re-configure products on the fly, e.g. a router goes down and its replacement doesn't have the exact capability so the service is impacted.  Resources could be logical, software, hardware etc.  Again, we had this in production in Nortel for its networks.

    Interestingly, we applied Neural Networks to analyse faults and found they weren't as good as humans looking at fault streams and understanding what the root cause/problem was. This was because although NNs are good at pattern analysis, they weren't good at spotting interwoven patterns but humans were.

    Much of this is still around online.

    Dave

    ------------------------------
    David Riches
    TM Forum
    Handyman
    +44 774 811 8071
    driches@tmforum.org
    ------------------------------



  • 4.  RE: YANG like model for physical equipment onboarding

    Posted Feb 04, 2021 07:43
    Thanks for that Dave,
    Interested in your opinions on if and how vendors might grant access to such capability to CSP to incorporate into their own internal processes.

    ------------------------------
    Paul Jordan
    BT Group plc
    ------------------------------



  • 5.  RE: YANG like model for physical equipment onboarding

    TM Forum Member
    Posted Feb 05, 2021 08:57
    Hi Paul,
    Not for me to comment on how vendors might do that, but certainly something you can ask as a customer.  Of course, I would expect BT Research Labs would have done something similar and when I was at Nortel there was a lot of collaboration going on on similar topics.  The concept of resource decomposition has been around for a longtime and fits nicely with the TM Forum's concept of Product/Service/Resource so I would expect there to be a good choice available in the industry.

    But then again, we though AI/ML etc. was an obvious choice back in the'90s. :-)

    Dave

    ------------------------------
    David Riches
    TM Forum
    Handyman
    +44 774 811 8071
    driches@tmforum.org
    ------------------------------



  • 6.  RE: YANG like model for physical equipment onboarding

    TM Forum Member
    Posted Feb 05, 2021 09:19
    Hi Saravana,
    I currently work in the physical compute/NFVI space, but in the past I've worked higher up in the stack, on the HSS and earlier in the OSS space. Netconf/Yang was a thing with the HSS and we implemented a Netconf/YANG interface for the management of the HSS.
    In the physical compute/storage/networking space I 've noticed that different protocols are in favor. DMTF's Redfish is most popular for Compute (and making inroads into the Networking space, formerly ruled by Netconf/YANG) while Swordfish (still developing) is the protocol-du-jour for Storage.
    If you haven't already, you might look at this: Redfish White Paper (dmtf.org)
    I believe most vendor's physical servers , some racks, some switches are redfish compliant, and provide an easy mechanism to traverse the physical equipment hierarchy.
    Here is a link to HPE's iLO Redfish API: API Reference (hewlettpackard.github.io)

    -----------------------
    Mario B Vincent
    HPE
    -----------------------

    ------------------------------
    Mario Vincent
    Hewlett-Packard Enterprise
    ------------------------------



  • 7.  RE: YANG like model for physical equipment onboarding

    Posted Feb 06, 2021 04:41
    Thank you Mario. Interesting. That looks like the kind of thing I was striving for.​

    ------------------------------
    Paul Jordan
    BT Group plc
    ------------------------------



  • 8.  RE: YANG like model for physical equipment onboarding

    TM Forum Member
    Posted Feb 08, 2021 09:58
    Here's some of the original work:
    • https://scienceon.kisti.re.kr/srch/selectPORSrchPatent.do?cn=USP2003076600749
    Which led me to find related work such as:
    • https://patents.google.com/patent/US8082335B2/en
    • https://patents.google.com/patent/US7941309B2/en
      • which talks about policy control mechanisms
    • https://patents.google.com/patent/US8380833B2/en

    One thing you might consider is using TM Forum's Catalyst program to investigate the challenge(s) you're looking at and see which delivery partners can help.

    If you'd like to find out more about this possibility then have a chat with your Engagement Manager, @Mike O'Sullivan

    Dave​​​

    ------------------------------
    David Riches
    TM Forum
    Handyman
    +44 774 811 8071
    driches@tmforum.org
    ------------------------------