Dear all, Please find below a WG item proposal for “Operation Automation for ALTO”, on which your feedback and suggestions will be more than welcome.
Thanks, Jensen ++++++++++++++++++++++++ Operation Automation for ALTO ============================= ## Proposed paragraph Best practice documentation and extension of ALTO for operation automation. Operating an ALTO server can be complex, and the automation of the operations of the ALTO services can be highly valuable. The operations of an ALTO server includes decisions on the set of information resources (e.g., what metrics, how they are divided into multiple entries) exposed in the information resource directory (IRD), population (may be proactive and reactive) of the content of the services (e.g., pull the backend, or trigger just-in-time measurements), and aggregation/processing of the collected information to give clients the proper response. Because of the dynamicity of network states and application demands, the automation of the ALTO server operations requires more inputs from applications. The working group will (1) investigate the best practices in ALTO operations automation, including interop/interfaces/protocols with routing systems and measurement tools; (2) propose the new protocol extension for more application-aware information to enable reactive measurement and computation. ## Basic Issue - New complexity of operations - Practical operations of an ALTO server includes the complex workflow - New ALTO protocol extensions (e.g., the incremental update mechanism, cost calendar, unified properties, and path vector) introduce more complexity to operations - Inefficient measurement and computation - ALTO information resource may be populated reactively, which requests more communications with applications to predict/decide which low-level information (e.g., INT) should be collected/measured, and which intermediate information (e.g., end-to-end path cache) should be precomputed ## Potential solution(s) - Investigate best practices (extending RFC7971) including: - Automatic IRD generation - ALTO information resources population from information collected by existing routing systems and measurement tools (e.g., TEDB, LSPDB, SNMP) - Reactive ALTO information resources update - Aggregation of different information sources - Considerations about intermediate abstraction (e.g., history, path cache) - Define the new interface (service) between application and server to allow the application to subscribe demand and enable reactive measurement and computation: - Interested E2E flows - Constraints for flows ## Remaining issues to be addressed - How to reactively populate on-demand ALTO information resources (e.g., path vectors) based on client queries - Flows that applications are interested in may change because of dynamic network state - e.g., applications are only interested in the endpoints attached to the specific PoP (Point of Presence) - Constraints for interested flows may be correlated - e.g., filter all sources whose total upload bandwidth are higher than the threshold ## Who will work on it - Luis/Jensen/Kai - Any other interested people ## Potential milestones (1-2 years) - IETF 109: problem statement - IETF 110: initial draft for protocol extension - IETF 112: best practice documentation ++++++++++++++++++++++++
_______________________________________________ alto mailing list alto@ietf.org https://www.ietf.org/mailman/listinfo/alto