Re: [Teas] TE Topology YANG Model: connectivity-matrix definition

[Igor Bryskin <[e-mail]>]

Hi Dieter,

One thing to keep in mind is that LTP X could be connected to LTPY with a metric (e.g. delay) M1, while to LTPZ with a different metric M2. This is especially important when a node is abstract (e.g. representing a domain) rather than a physical switch. We might think about equal cost single list optimization you are proposing, but this is just an optimization, which is even less important when CM changes very rarely (as the case with physical switches). Abstract node CM may change dynamically, but it is unlikely to contain equal cost entries.
This is to explain as to why we modeled CM the way we did.

Igor

From: Dieter Beller [mailto:[e-mail]]
Sent: Friday, July 21, 2017 8:31 AM
To: Xufeng Liu; Vishnu Pavan Beeram; Igor Bryskin; Oscar Gonzalez De Dios; Tarek Saad; Himanshu Shah; Lou Berger; BRUNGARD, DEBORAH A (ATTLABS); Susan Hares; Zafar Ali (zali); Khaddam, Mazen (CCI-Atlanta); Tony Le; BELOTTI, SERGIO (SERGIO); Beller, Dieter (Dieter); Rajan Rao; Zhangxian (Xian); [e-mail]; Belotti, Sergio (Nokia - IT); Anurag Sharma
Cc: [e-mail]
Subject: TE Topology YANG Model: connectivity-matrix definition

Hi Xufeng, Igor, all,

looking again at the connectivity-matrix definition, I have the following question: based on the current definition there are multiple separate entries
for the same "from TP" to all potential "to TPs":

   +--rw connectivity-matrix* [id]

      +--rw id                          uint32

      +--rw from

      |  +--rw tp-ref?              leafref

      |  +--rw label-restriction* [inclusive-exclusive label-start]

      |     +--rw inclusive-exclusive    enumeration

      |     +--rw label-start            rt-types:generalized-label

      |     +--rw label-end?             rt-types:generalized-label

      |     +--rw range-bitmap?          binary

      +--rw to

      |  +--rw tp-ref?              leafref

      |  +--rw label-restriction* [inclusive-exclusive label-start]

      |     +--rw inclusive-exclusive    enumeration

      |     +--rw label-start            rt-types:generalized-label

      |     +--rw label-end?             rt-types:generalized-label

      |     +--rw range-bitmap?          binary

I think that it is more efficient to have a single entry for each "from TP" and a list with all potentially connectable "to TPs"

      +--rw id   uint32

      +--rw from

      |  +--rw tp-ref?   leafref

      |  +--rw label-restriction* [inclusive-exclusive label-start]

      |     +--rw inclusive-exclusive   enumeration

      |     +--rw label-start           rt-types:generalized-label

      |     +--rw label-end?            rt-types:generalized-label

      |     +--rw range-bitmap?         binary

      +--rw to-tp-list* [tp-id]

      |  +--rw tp-id   te-types:te-tp-id

      |  +--rw tp-ref?   leafref

      |  +--rw label-restriction* [inclusive-exclusive label-start]

      |     +--rw inclusive-exclusive   enumeration

      |     +--rw label-start           rt-types:generalized-label

      |     +--rw label-end?            rt-types:generalized-label

      |     +--rw range-bitmap?         binary


If the label restrictions of the "from TP" are the same (or do not exist, which is typically the case) for multiple or all connectable "to TPs", the
connectivity matrix description using the to-tp-list becomes much more compact compared to the description above. (I am not sure whether
the tp-id can be used as list index. So, if this is not correct, it has to be replace it by some other index.)


Thanks,
Dieter