Markdown Version | Session Recording

Session Date/Time: 20 Feb 2025 13:00

ANIMA

Summary

This was the first ANIMA interim meeting of 2024, convened to discuss a series of proposals for new work within the working group. The primary goal was to assess how these proposals align with the existing ANIMA charter and the spirit of its work, particularly focusing on the use of Autonomic Service Agents (ASAs) and the ANIMA framework (ACP, BRUU, GRASP) for in-network automation. Discussions centered on identifying standardizable components, potential extensions to existing work, and the importance of practical implementations and proof-of-concept for proposed solutions.

Key Discussion Points

The chairs provided an overview of the ANIMA working group's focus on operational standardization within network automation, emphasizing the role of Autonomic Service Agents (ASAs) running on network devices atop the ANIMA infrastructure (ACP for connectivity, BRUU for mutual trust, GRASP for quick protocol building). Key aspects for standardization include ASA interactions and procedures. The importance of specifications that are implementable and make sense to operators, with proof-of-concepts, was highlighted.

  1. Autonomic Distributed Troubleshooting (Presented by Leo Liu)

    • Problem: Traditional troubleshooting methods (manual CLI, centralized NMS/AI, protocol-specific mechanisms) are time-consuming, resource-intensive, or lack generalization.
    • Proposal: Generalize the protocol-specific approach using the ANIMA platform, specifically GRASP, as an independent troubleshooting tool.
    • Use Cases: Routing-related issues (e.g., BGP route flapping source identification, MPLS path failures) and general management protocol failures (e.g., SNMP/NETCONF failure).
    • ANIMA/GRASP Fit: GRASP as a signaling protocol to carry OAM information between devices.
    • Requirements: New GRASP options for troubleshooting info, reuse existing GRASP messages/interactions, lightweight implementation (UDP preferred), decoupling from ACP (optional, possibly DTS or GRASP security extensions).
    • Discussion:
      • Feedback requested more detailed examples, especially for a single routing protocol, focusing on specific ASA actions, state tracking, problem discovery, peer communication, and resolution.
      • Clarification that the focus is on operational/management mechanisms rather than extending routing protocols.
      • A participant noted ongoing work on lightweight GRASP based on CoAP, which could be relevant.
      • The decoupling from ACP was discussed; while ACP offers strong security, its IPv6 dependency and setup complexity can be barriers in some small/medium enterprises, but it should remain an option.

  2. Home Broadband Automatic Installation and Maintenance (Presented by Huaming Zhan)

    • Problem: Home broadband installation and maintenance rely on human experience, leading to variable efficiency and untimely information transmission.
    • Proposal: Use intelligent agents combined with Large Language Models (LLM) and API orchestration for intelligent diagnosis and repair, improving efficiency.
    • ANIMA/GRASP Fit: GRASP protocol for collaborative innovation, enhancing efficiency through self-service intelligent solutions. GRASP message types (Discovery, Negotiation, Synchronization, Flooding) mapped to a 10-step process involving LLM, agent, and capability providers.
    • Requirements: Optimize GRASP for device discovery (adjust timeouts) and negotiation (high concurrency, multi-threading/event loop mechanisms).
    • Discussion:
      • Difficulties with slide sharing and reading led to limited detailed technical discussion during the presentation.
      • Feedback requested more clarity on how GRASP specifically enables the proposed mechanism and its interaction with LLMs and agents.

  3. Application-aware Routing for Enhanced Services (Presented by Feng Guo)

    • Problem: SRv6 services are not directly visible or controllable by end-users, and current QoS/SLA mechanisms cannot meet differentiated service requirements.
    • Proposal: Introduce Application-aware Routing ID (AR ID) to make SR services visible, allowing steering of user traffic to specific SR policies or network slices on demand. AR IDs have user and network types, mapped to SR policies/slices.
    • ANIMA/GRASP Fit: GRASP for cross-domain negotiation of AR IDs, especially in complex scenarios where optimal path selection requires multiple rounds of negotiation, simplifying coordination compared to centralized controllers.
    • Discussion:
      • The proposal's relationship to the ongoing Naser (Network Slice and Service Routing) initiative was raised; its scope might be broader than ANIMA's typical focus.
      • Clarified that GRASP's role would be primarily for negotiation across domain edges, as intra-domain control might already be handled by existing controllers.
      • Feedback suggested to consider the complexity and scope in relation to other IETF WGs (e.g., Routing Area) and potentially narrow the ANIMA specific aspects.

  4. Autonomic Network Congestion Relief (Presented by Jianan Yuan)

    • Problem: Fiber optical failures cause network congestion, requiring manual inspection and configuration adjustments, increasing operational burden.
    • Proposal: An autonomic congestion relief mechanism based on intelligent traffic analysis and auto-regulation.
    • Mechanism: Intelligent modules on devices perform traffic modeling (full flow data, top-N identification), monitoring (BGP-LS extension for link bandwidth/load), intelligent policy generation (redistribute traffic), and policy propagation/revision (BGP RPD protocol, adjusting routing priorities).
    • ANIMA/GRASP Fit: GRASP was indicated as a potential future consideration for negotiation and control activities.
    • Discussion:
      • Feedback requested concrete examples of policies applied before and after failure detection, detailing how they differ and how GRASP messages would be used for negotiation.
      • Recommendation to subscribe to the ANIMA mailing list for further discussion and to provide suggestions on drafting.

  5. Mobility Management and Capability Negotiation (Presented by Yang Yang)

    • Problem: When mobile nodes access new networks or hand over, selecting the appropriate mobility management protocol (host-based vs. network-based, with various extensions) requires negotiation between the host and network.
    • Proposal: Standardize principles for protocol selection (network capability, host performance, functional/performance enhancement) and use ANIMA protocols for capability notification and negotiation.
    • ANIMA/GRASP Fit: GRASP or other ANIMA protocols to notify and negotiate mobility capabilities and parameter settings between host and network.
    • Discussion:
      • Feedback requested more concrete and detailed examples of the negotiation process, especially concerning the information elements to be signaled.
      • Inquired about existing prototypes or practical work.
      • Distinction between GRASP and ICMPv6 (used in prior drafts) was highlighted, suggesting to clarify communication patterns (point-to-point, flooding, discovery) to justify GRASP.
      • Clarified that negotiation occurs during initial bootstrap/access and during Handover to a new network.

  6. Autonomic Campus Network Bootstrapping and VLAN Auto-Configuration (Presented by Leo Liu)

    • Problem: Campus networks (especially small/medium enterprises) require automated bootstrapping and VLAN configuration without manual intervention, dedicated controllers, or complex L3 routing. Resource-constrained devices are common.
    • Proposal: Utilize GRASP as a "glue" for discovery, rule definition (AP, switch, gateway roles), and self-selection of a "local controller" (e.g., on a gateway device). This local controller uses GRASP to deliver compiled VLAN configurations to relevant devices, replacing traditional management protocols.
    • ANIMA/GRASP Fit: GRASP for device discovery, L2 topology collection, local controller election, and configuration delivery.
    • Requirements: New GRASP options for bootstrapping/VLAN configuration, reuse existing GRASP messages/interactions, lightweight (UDP preferred), ACP optional.
    • Discussion:
      • Comparison to existing proprietary (Cisco VTP, Extreme) and IEEE (GVRP) VLAN signaling protocols was requested, asking for clear distinctions or new requirements that an IETF solution would address beyond merely standardizing existing functionality.
      • Clarification on the "no controller" assumption vs. the introduction of a "local controller" (which acts as a logical/virtual controller integrated into a device like a gateway, distinct from a dedicated system).
      • Suggestion to focus on solving additional deployment requirements rather than just adding "new features."

  7. Naser Use Case for ANIMA (Presented by Peter Huang)

    • Problem: Current traffic security is often limited to encryption/signing; clients require enhanced physical security (e.g., data staying within a country) and specific trust properties (crypto algorithms, integrity checks). Centralized path computation is a bottleneck.
    • Proposal: Provide a framework for operators to offer enhanced services by picking trusted devices, orchestrating secure paths, and verifying compliance. This requires distributed trust evaluation and trusted path setup to avoid controller bottlenecks.
    • ANIMA/GRASP Fit: ANIMA could provide autodiscovery of trusted devices, distributed trust evaluation (attestation results shared peer-to-peer), and distributed agreement for trusted path setup, leveraging GRASP for signaling.
    • Discussion:
      • The relationship to the broader Naser initiative (currently in the security area) was discussed, noting that while Naser might focus on attesting paths, the path computation and creation of trusted paths might fall to other WGs.
      • The impact on compute resources for trusted path calculation was discussed, with a sense that it might not be a significant burden.
      • GRASP was suggested as a suitable framework for signaling additional information needed for Naser, offering secure point-to-point and group communication via flooding/discovery.

Decisions and Action Items

Next Steps