• Host: metal-01.he-eu-hel1.vacdst.misc Resolution: 30s Nodes: 300

• Host: metal-01.he-eu-hel1.vacdst.misc, metal-01.he-eu-fsn1.vacdst.misc Resolution: 30s Nodes: 300 (different datacenter)

• Host: metal-01.he-eu-fsn1.vacdst.misc, metal-02.he-eu-hel1.vacdst.misc Resolution: 30s Nodes: 300 (different datacenter)

• Host: metal-01.he-eu-hel1.vacdst.misc, metal-02.he-eu-hel1.vacdst.misc Resolution: 30s Nodes: 300 (same datacenter)

• Host: metal-01.he-eu-hel1.vacdst.misc, metal-01.he-eu-fsn1.vacdst.misc, metal-02.he-eu-hel1.vacdst.misc Resolution: 30s Nodes: 300

Notes:

• Using only one host, bandwidth usage is around 15MB/s. Two hosts makes the bandwidth increases to 18MB/s~. 3 hosts makes the bandwidth increase to 20MB/s.
• Packet drop looks like is higher when 2 hosts are in same datacenter (metal-01 vs metal-2 in he-eu-hel1).
• Dropped packets increases drastically when using 3 nodes.

Changelog:

• Better adjustment of gossipsub parameters with nWaku:
  • Heartbeat (700ms to 1 second)
  • Prunebackoff (3 seconds to 60 seconds)
  • gossipFactor (0.05 to 0.25)

Libp2p changes:

Notes