Update loopback-mode type

[sallylsy]

Change Scope

• In openconfig-transport-types.yang, updates the current enums of loopback-mode-type.
• Two of the existing enums are updated to new values. If users were using FACILITY or TERMINAL, they should update to ASIC_MAC_LOCAL/PHY_MAC_LOCAL or ASIC_MAC_REMOTE/PHY_MAC_REMOTE

Ref:

Reference to SAI implementation

SAI header

Arista EOS - traffic loopback supports local and remote options. It also allows specifying MAC and PHY layer options.

JunOS - set loopback supports local loopback only. JunOS does not expose a MAC or PHY level option from their configuration

Cisco IOS XR - loopback supports local (aka terminal or internal) and remote (aka facility or line) loopbacks. Cisco does not expose a MAC or PHY level option from their configuration.

[OpenConfigBot]

No major YANG version changes in commit 2eb5cd6

[dplore]

We will review in OC operators meeting on Jan 23, 2024

[dplore]

I am not a deep expert on transceiver loopbacks, but I thought some transceiver modules implement a PHY level loopback in the transceiver module (as opposed to the ASIC). Is this true?

@jnelson-sj do you have experience here? Does it make sense to have some loopback enum for a transceiver PHY level loopback?

[jnelson-sj]

+Aditya Bhiday ***@***.***> Hi Darren, Yes, there are transceivers which perform loopbacks at different levels of the processing chain. For example, ZR transceivers have the following loopbacks defined in OIF-400GZR: [image: image.png] You can (hopefully) see in the diagram that there are 6 loopbacks defined here. The can be grouped into whether they provide a FACILITY or a TERMINAL loopback: FACILITY Loopbacks: - Media Side Tx - Tx Modem - Host Side Tx TERMINAL Loopbacks: - Media Side Rx - Rx Modem - Host Side Rx I think the PHY level loopback you are talking about may be the Host Side Tx - what the switch port sends on its tx is looped back to its rx with minimal processing in the transceiver. I wonder if there should be another dimension to the loopbacks - a depth. The FACILITY/TERMINAL gives the direction of the loopback, but it would also be good to have a DEEP/SHALLOW qualifier. This would be used to specify how much of the transceiver function is to be exercised in the loopback. Is this along the lines of what you are thinking? John

…

On Tue, Jan 23, 2024 at 9:17 AM Darren Loher ***@***.***> wrote: I am not a deep expert on transceiver loopbacks, but I thought some transceiver modules implement a PHY level loopback in the transceiver module (as opposed to the ASIC). Is this true? @jnelson-sj <https://github.com/jnelson-sj> do you have experience here? Does it make sense to have some loopback enum for a transceiver PHY level loopback? — Reply to this email directly, view it on GitHub <#1034 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/A3WYQG2ZFYJS66F6PPQIRHTYP7WCLAVCNFSM6AAAAABB5WZUJOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSMBWGU2DSMBZGY> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[dplore]

Hi John, thank you for the insight! My bias is to handle the 400ZR case separately as I think the ASIC MAC/PHY layer loopbacks proposed here are still applicable and do not appear to conflict with the 400ZR case.

+Aditya Bhiday @.***> Hi Darren, Yes, there are transceivers which perform loopbacks at different levels of the processing chain. For example, ZR transceivers have the following loopbacks defined in OIF-400GZR: [image: image.png] You can (hopefully) see in the diagram that there are 6 loopbacks defined here. The can be grouped into whether they provide a FACILITY or a TERMINAL loopback: FACILITY Loopbacks: - Media Side Tx - Tx Modem - Host Side Tx TERMINAL Loopbacks: - Media Side Rx - Rx Modem - Host Side Rx I think the PHY level loopback you are talking about may be the Host Side Tx - what the switch port sends on its tx is looped back to its rx with minimal processing in the transceiver. I wonder if there should be another dimension to the loopbacks - a depth. The FACILITY/TERMINAL gives the direction of the loopback, but it would also be good to have a DEEP/SHALLOW qualifier. This would be used to specify how much of the transceiver function is to be exercised in the loopback. Is this along the lines of what you are thinking? John
…

[dplore]

This was reviewed in the Jan 30, 2024 OC Operators meeting. The feedback was we should address how a system which does not expose MAC / PHY layer loopbacks, but only an 'interface' loopback should be handled. The suggestion was to
retain the existing FACILITY and TERMINAL enum values rather than deprecating them.

[dplore]

LGTM. Last call for comments please. This will merge on Feb 13, 2024

[dplore]

/gcbrun

[dplore]

/gcbrun