Markdown Version | Session Recording

Session Date/Time: 05 Nov 2024 16:30

srv6ops

Summary

The srv6ops working group meeting focused on operational experiences with SRv6 deployments across various sectors, including telecommunications, power grids, and enterprise IT. Four presentations were given, covering service programming implementation, SRv6 deployment in a power grid network, network analytics for SRv6 incidents, and SRv6 use in Verizon's IT data centers. The meeting highlighted the benefits, challenges, and solutions encountered by operators in different deployment scenarios.

Key Discussion Points

• Bell Canada's Service Programming Implementation (Daniel):
  • Focus on creating telco service constructs (subscriber session steering, service chaining, traffic engineering) using SRv6.
  • Emphasized stateless traffic steering using SR policies and a proxy function.
  • Achieved significant cost reduction compared to traditional methods.
  • Discussion on using a custom SID for service chaining.
• China Southern Power Grid's SRv6 Deployment (Lu Jiang):
  • Motivation for SRv6 deployment driven by new digital services and limitations of MPLS.
  • Discussed a migration strategy using MPLS/SRv6 interworking to modernize dispatch backbone network incrementally.
  • Utilized SRv6 policies for load balancing video surveillance traffic.
  • Integrated SRv6 with an SDN platform for path optimization.
• Swisscom's SRv6 Network Incident Analytics (Thomas):
  • Presentation on using network telemetry and anomaly detection to identify and analyze network incidents.
  • Described the collection of metrics from the network and integration into a data mesh.
  • Showed how anomaly detection helped identify the root cause of application-level issues during a maintenance window.
  • Highlighted the need for auto-profiling and replay capabilities for anomaly detection.
• Verizon IT Data Centers SRv6 Use (Gyan Mishra):
  • Evolved from a large Layer 2 domain to a multi-pod SRv6 architecture.
  • Shared operational considerations encountered in inter-domain routing.
  • Emphasized the importance of "next hop unchanged" configuration on EBGP peering for proper service SID propagation (RFC 9252).
  • Discussed host integration with SRV6

Decisions and Action Items

• Action Item: Daniel (Bell Canada) to contribute a document to the working group and provide guidance based on their SRv6 experience.
• Action Item: Gyan (Verizon) to further refine the presentation content and consider if it can be used as a document for the working group. Discuss with chairs if the document would better describe a deployment scenario than an operational guidance.
• Action Item: Discuss Traffic Engineering concerns in Data Centers on the mailing list.
• Action Item: Thomas (Swisscom) to potentially provide input on the charter item regarding network observability based on their experiences.

Next Steps

• Continue discussion on the mailing list regarding traffic engineering in data centers, operational guidance vs deployment scenario.
• Encourage Daniel to contribute a document to the working group on SRv6 service programming.
• Follow up with Thomas regarding network observability input.
• Prioritize documents for discussion in future meetings.