Lab and simulations have got more stable. This can be seen in the similarities between in and out bandwidth with respect to the same experiments. Comparing with our previous report (Waku first deliverable - Feb 2024 ), there are a couple of improvements that can be seen:
In v0.29, the bandwidth usage has been increased by roughly a ~20%. This simulation has been repeated twice because of a secondary issue, but the results were very similar.
Note: The only difference we had to do on v0.29 due to issue https://github.com/waku-org/nwaku/issues/2780 was increasing the number of maximum connections (200 to 500 in bootstrap nodes, 50 to 150 in normal nodes). Unless metadata connections could make this increase, it is something that might be worth to investigate.
Median libp2p bandwidth usage per node in the simulation
Versions: v0.27 v0.28 and v0.29
Mesh size: 1000, 2000 and 3000 nwaku nodes
Message rates: 1msg/s, 1msg/5s, 1msg/10s
Message size: 1KB
Issues found fixed: