> But isn't this exactly where you would either start copying same data N times > if more then one app want to include given metric in its topo computation ?
If more than one app wants to use the same metric, then there is no need to copy anything. You simply all refer to the same metric. It’s a flat space. > Or alternatively we would quickly get into mess of overlapping pools of > metrics used by app A and app B but not app C. I don't think troubleshooting > the network like this would be fun. I don’t see that as a mess. If you don’t like overlap, you don’t have to. Again, there are lots of metrics. 256 of them. If you burn one per FAD, you’ve still got 128 left for future applications. > Consistent configuration on all nodes to use the same metrics for a given app > is also a challenge, but I guess FAD server push or general central config > push could help here. It’s 2021. Consistent configuration is driven by management plane automation. Yes, consistent FAD definitions are achieved by (surprise!) FAD advertisement. :-) Tony _______________________________________________ Lsr mailing list Lsr@ietf.org https://www.ietf.org/mailman/listinfo/lsr
