Analysis: Waku Relay scaling (part 1)

vacp2p / research

Thinking in code

MIT License

62 stars 4 forks source link

Analysis: Waku Relay scaling (part 1) #158

Closed kaiserd closed 1 year ago

kaiserd commented 1 year ago

Anaylsis based on simplified assumptions: How does a single pubsub mesh scale, including an outlook towards what topic sharding can achieve.

corpetty commented 1 year ago

tagging Distributed Systems Testing team: @AlbertoSoutullo, @0xFugue, @Daimakaimura

oskarth commented 1 year ago

I'd just use this python file as a starting point https://github.com/vacp2p/research/tree/master/whisper_scalability and update code to make sure we capture relevant dimensions

kaiserd commented 1 year ago

Yes. My plan was to do a napkin calculation first, and think about relevant dimensions. After that, adjusting the python script. (Plan is to get first results here this week.)

oskarth commented 1 year ago

@rymnc or @alrevuelta might be able to help to with model stuff depending on how things go with other pieces of work (RLN/Community specs and peer connection management) and how parallelizable this issue would be

kaiserd commented 1 year ago

here is the analyis. It also contains sharding.
here is the python script used to the generate the results and plots.

This concludes the 1st part of the Analysis. Future work on this will be tracked in a new issue.

AlbertoSoutullo commented 1 year ago

Hi, how can we have access to that?