GNA-G Routing WG
Network owners and users need to ensure data is moving over the research and education (R&E) circuits correctly so that data transfers are performing well. Experience has shown that adding or removing capacity can have unexpected routing results that are difficult to detect automatically and impossible to correct without coordination. The anomalous flows can be identified fairly easily with current tools, but it is still challenging to work across the R&E networking community to adjust the erroneous paths, which can be complicated due to the overall number of organizations involved. However, this type of routing problem is prevalent and growing as more capacity along different routes is added, and we expect this issue is one many NRENs will need to address in order to maintain a robust, reliable, high-speed global R&E network.
Some of the problems we currently see with R&E data transfers and routes include:
- Data taking a longer route than necessary, for example, unnecessarily crossing oceans.
- Traffic taking an unexpected route, for example, hitting two routers in a single exchange point, before and after passing through a third (likely unnecessary) exchange point.
- Traffic being routed over commercial capacity instead of remaining on R&E capacity.
The GNA-G, and this WG in particular, provide a natural context for this work, given the inter-regional extent of these issues that affect facilities and sites located in several regions of the world, and the essential need for common cooperation to resolve them.
This working group will engage network owners and NRENs to not only reactively discuss and address ineffective routes, but will work proactively across the community to systematically create policies to prevent them from occurring.
Fun Fact: We don’t use the word optimal in this work because “optimal” is a mathematical concept and not achievable in practice. There are many reasons routes are done certain ways, and because of that, the goal of this group must be to try to have effective routes that follow a known policy, which may or may not be the most efficient or the most mathematically pleasing.