Open italobusi opened 4 years ago
I think we should consider the following cases:
- symmetric constraints ----- symmetric properties
/ \
/ \
bidir path - \
\ \
\- asymmetric constraints ----- asymmetric properties
The model has provision for reverse-path-properties for bidirectional. The reverse path properties are displayed under the reverse-path.
The reverse-path is configured for non-corouted case The reverse-path is not configured for co-routed. The properties for computed path can always be displayed under the reverse path
The forward path has constraints and properties If the reverse path is configured, then the bidirectional tunnel "can" be asymmetric and non-co-routed. The reverse path properties can be specified differently from the forward path
note: for asymmetric BW can be specified different in forward/reverse
See also discussions in open issues #25 and #83
Co-routed:
Open issue for secondary. AI(Italo): to produce example and highlight issue
Some slides describing my doubts: tunnel-paths-01.pptx
Main questions:
Slides updated based on the 2021-02-12 discussion: tunnel-paths-03.pptx
Some slides describing my doubts: tunnel-paths-01.pptx
Main questions:
- How to know which secondary-reverse path is co-routed with which secondary-path: see slide 4
Use the order in the Candidate-Secondary-list and in the Rev-Candidate-Secondary-list: the nth element of the Rev-Candidate-Secondary-list is co-routed with the nth element in the Candidate-Secondary-list (see updated slide 4)
- When constraints are configured only for the forward path, how to distinguish between the configuration of a bidirectional tunnel with symmetric constraints requesting symmetric properties or asymmetric properties: see slide 5
When requesting asymmetric properties, the reverse path are configured:
JSON code for the examples in tunnel-paths-03.pptx: tunnel-paths-03-00.zip
Added node identifiers (following RFC 5737 requirements): tunnel-paths-04-00.zip
See also PR https://github.com/tsaad-dev/te/pull/191
The topology in A.1 is not sufficient to be a reference for these examples. Should we define a new Appendix or change the topology reference in A.1?
I have noted that the examples were not aligned with the new co-routed leaf defined during WG LC comments resolution. The examples have been updated accordingly: tunnel-paths-05.zip
I have few questions/doubts:
when
statement in the YANG code shall be removed or updated (see the examples in slides 3 and 4)Pending resolution of #203
Pending also resolution of #214 (which in turns depends on #203)
I do not remember why it has been assigned only to the Tutorial
IMHO, we need to resolve the comments above and fix also the JSON example in section 12.6 of the TE tunnel draft
bring back preference to secondary-reverse and add when does preference kick in between multiple paths. Specifically, when multiple paths protect against the same failure (for secondary paths).
bring back preference to secondary-reverse and add when does preference kick in between multiple paths. Specifically, when multiple paths protect against the same failure (for secondary paths).
Copied to the right issue: https://github.com/tsaad-dev/te/issues/192#issuecomment-1453805571
Example in slide 1 updated as requested in https://github.com/tsaad-dev/te/issues/192#issuecomment-1453767703: tunnel-paths-06.pptx
Should we update the JSON example accordingly?
Do we need to provide some text description of these examples in section 12.6?
The description "Example in slide X" would not be clear to a reader who is not familiar with these slides ...
Example in slide 1 updated as requested in #192 (comment): tunnel-paths-06.pptx
Should we update the JSON example accordingly?
Do we need to provide some text description of these examples in section 12.6?
The description "Example in slide X" would not be clear to a reader who is not familiar with these slides ...
Pending resolution of issue #192 : align the slides the the latest version in #192
Assigned to Italo. No action on the TE tunnel. Should not be blocking on progressing the TE tunnel to LC
The module misses a mechanism to include a metric's accumulative-value per direction, e.g. we could have a bidirectional WDM LSP with 2 different SNR values, one for downstream and one for downstream; a possible solution could be to add an optional "upstream" branch carrying relevant properties.
See: https://github.com/rvilalta/ietf-te-path-computation/issues/65