ECMP and LAG hashing logic
Posted: 12 Feb 2018 17:35
Hello,
I'm having issues understanding the hashing options and expected behavior of load-balancing in different scenarios, especially within services.
Are there any specific show commands that show hashing results or logic that is currently applied within a context like global, within different service types, across LAGs, across multiple LSPs on a given spoke SDP in different services?
I have pored through documentation and am building a picture, but I am also not sure how deep within a packet the SR will inspect for hashing information when hashing is being applied, or how many times it's done and the order of operations. This may be a fundamental misunderstanding on my own part with the separation of contexts within a device or per-hop, and what "type" of traffic each flow is considered once its encapsulated with multiple tags.
I would like to have the tools to build this picture for myself, or to understand the tools I have available a bit better.
For instance, if I have a LAG that is a SAP in a VPLS, and each spoke/mesh-SDP has multiple LSPs using unique paths... you get the idea. I need to be able to built a logic table for each step of hashing with different service types, traffic types, link types, etc.
Any ideas on correcting my perspective, or helping me visualize this a different way? Any help is much appreciated.
I'm having issues understanding the hashing options and expected behavior of load-balancing in different scenarios, especially within services.
Are there any specific show commands that show hashing results or logic that is currently applied within a context like global, within different service types, across LAGs, across multiple LSPs on a given spoke SDP in different services?
I have pored through documentation and am building a picture, but I am also not sure how deep within a packet the SR will inspect for hashing information when hashing is being applied, or how many times it's done and the order of operations. This may be a fundamental misunderstanding on my own part with the separation of contexts within a device or per-hop, and what "type" of traffic each flow is considered once its encapsulated with multiple tags.
I would like to have the tools to build this picture for myself, or to understand the tools I have available a bit better.
For instance, if I have a LAG that is a SAP in a VPLS, and each spoke/mesh-SDP has multiple LSPs using unique paths... you get the idea. I need to be able to built a logic table for each step of hashing with different service types, traffic types, link types, etc.
Any ideas on correcting my perspective, or helping me visualize this a different way? Any help is much appreciated.