The throughput of discovery (e.g., 20 peers/sec) is not enough, we need quality metrics. Ideally, the newly discovered nodes for a topic should be disjoint from existing (already discovered) nodes and their distribution should be uniform.
Also, a metric to measure "time-to-full-discovery" can be useful: if there are 100 nodes out of 10K that are advertising a topic, how long does it take to discover all of the 100 nodes.
This applies to topic discovery.
The throughput of discovery (e.g., 20 peers/sec) is not enough, we need quality metrics. Ideally, the newly discovered nodes for a topic should be disjoint from existing (already discovered) nodes and their distribution should be uniform.
Also, a metric to measure "time-to-full-discovery" can be useful: if there are 100 nodes out of 10K that are advertising a topic, how long does it take to discover all of the 100 nodes.