We have put together a new IPFS measurement report for calendar week 10 in 2023 from
2023-03-20 . You can find it here:
- first data from our revised website monitoring infra reporting TTFB, first/paint contentful paint as opposed to DOMContentLoaded. Reasoning here
- Based on feedback added an “Error Rate” chart to our DHT lookup performance measurements here
- DHT performance slightly improved compared to previous week
One other notable thing is that in the “kubo vs HTTP” comparison plot for websites, this week we see several websites being faster with kubo. It would be interesting to dig deeper to figure out what’s helping here. Is it larger number of providers for these websites, perhaps? Something to keep in mind to do for later.
Also, @dennis-tra, what is the definition of the error rate for publications and retrievals given here: network-measurements/README.md at master · protocol/network-measurements · GitHub ? Is it non-responsiveness after a pre-defined period of time? A 404 error? It’s worth clarifying as part of the report (i.e., adding to the template).
on the retrieval side it’s “provider record not found after 1m” and on the publication side it’s “provide operation timed out after 3m”.