Thunderdome: Run Initial Kubo Experiments

[iand]

What Is It?

Create and run some initial experiments in Thunderdome against live traffic.

Deliverables

• Docker images and Thunderdome configuration for each experiment.
• A Grafana dashboard and a short analysis of key results from running each experiment.

Why Are We Doing It?

Running these initial experiments will demonstrate that the basic Thunderdome infrastructure works and can produce interesting results. Some of the experiments have current practical value:

• Comparing the current release candidate of Kubo against the previous version could detect possible performance regressions or improvements.
• Comparing different values of the provider delay setting can inform the decision on what will be the best value to use by default in Kubo.
• Comparing different tracing levels can help us understand the trade off between performance and utility when enabling tracing in production

Notes

Experiments should run for a significant time such as 12 or 24 hours.

Tasks

• [x] https://github.com/ipfs-shipyard/thunderdome/issues/3
• [x] https://github.com/ipfs-shipyard/thunderdome/issues/9
• [x] https://github.com/ipfs-shipyard/thunderdome/issues/11
• [x] https://github.com/ipfs-shipyard/thunderdome/issues/33

[iand]

See https://www.notion.so/pl-strflt/Thunderdome-Update-7-Sep-2022-9679a762b59f423b87bc2258a9c7e1ec for summary of initial experiment runs