Dear Chongfeng, Thanks a lot for the review and the questions.
See below inline my reply. Apologies for late response. We intend to update the document with an implementation status section and example data template section soon. Best wishes Thomas From: [email protected] <[email protected]> Sent: Tuesday, December 30, 2025 10:56 AM To: Graf Thomas, SCS-INI-NET-VNC-E2E <[email protected]> Cc: opsawg <[email protected]> Subject: Comments on draft-ietf-opsawg-ipfix-gpon-gem Be aware: This is an external email. Hi Thomas, I have reviewed draft-ietf-opsawg-ipfix-gpon-gem, and have the following comments, This draft mainly use information such as the PTI and GEM PORT-ID in the GPON protocol frame, integrating them into IPFIX to distinguish between service flows and management flows like OMCI. However, I am not very clear about the value of this approach and have the following questions: 1. In the GPON protocol frame, PTI is used to indicate whether a frame is a service frame or an OAM frame, and it also serves as a fragmentation identifier within the GPON protocol. However, this identifier was deprecated in the subsequent 10G-PON, how do you consider this discontinunity? . TG> To my understanding, the PTI field is part of the GEM (GPON Encapsulation Method) header. The ITU-T standards that utilize the PTI include: BPON (Broadband PON), specified in ITU-T G.983. GPON (Gigabit PON), specified in ITU-T G.984 series (e.g., G.984.3). XGS-PON (10-Gigabit-capable Symmetric PON), specified in ITU-T G.9807.1. Subsequent generations like 50G-PON (ITU-T G.9805/G.9804 series) also extend the GEM encapsulation method, and thus incorporate the PTI field 1. The OAM frames identified in the GPON protocol essentially represent OMCI management information. However, the transmission and processing of this management information are confined to the GPON protocol stack software between the OLT and ONU. Therefore, if such information is placed into IPFIX, it is unclear whether it can be utilized for subsequent network transmission phase. Could you elaborate on the value of doing so? TG> Correct the OAM frames are only forwarded within the passive optical network between the OLT and ONU. Today, IPFIX is used in the passive optical network between the OLT and ONU to export the encapsulated ethernet properties. It misses the PON context. Here an example data-template of an existing implementation. IE1 octetDeltaCount IE2 packetDeltaCount IE3 deltaFlowCount IE10 ingressInterface IE14 egressInterface IE21 flowEndSysUpTime IE22 flowStartSysUpTime IE56 sourceMacAddress IE80 destinationMacAddress IE89 forwardingStatus TG> Additionally the following IE's would be added TBD1 gponGemPti TBD2 gponGemPortId TG> Of particular interest is the transition on the OLT from the passive optical network to the upstream ethernet domain. See slide 6 in https://datatracker.ietf.org/meeting/123/materials/slides-123-opsawg-export-of-gigabit-passive-optical-network-encapsulation-mode-in-ipfix-00. Network operators usually segment type of traffic among different GEM Port ID's. As example separating voice, video or Internet best effort. At Swisscom for example multicast TV is carried over a different GEM Port ID than Internet best effort. Or in case of mobile backholing, to support network slicing, for each slice a separate GEM Port ID and QoS class on Ethernet/IP is allocated. Best regards Chongfeng
smime.p7s
Description: S/MIME Cryptographic Signature
_______________________________________________ OPSAWG mailing list -- [email protected] To unsubscribe send an email to [email protected]
