Thinking about workshops for the following months, I would love to know if #participants:indexers would be interested in having an Indexer Performance Workshop, and what topics it should include
These are some of the specific questions i had about performance :
- How to identify bottlenecks in the overall indexer setup ?
- How to speed up indexing ?
- What’s the best performance metrics ?
- Is “blocks processing” an accurate performance indicator ? Why is it just below 120 on OpenEthereum and just below 60 on TurboGeth, bizarly a factor 2 ? While simple 5 min interval measurements of how many blocks are indexed, shows them much closer ?
- How to asses a setup’s capacity for handling real-world queries and which tweaks / improvements can be made to increase capacity ?
- Does indexing more than one subgraph at the same time over the same infrastructure (indexer, ETH archive) slow down the speed at which it happens ?
I’d like to see a workshop that covers the lessons learned on hosted. It could be broken up by indexing component:
- INDEXING: Optimising ethnode -> graph-node performance
- INDEXING: Optimising graph-node -> database performance
- SERVING QUERIES: Optimising query-node -> database performance
- OPS: Top metrics monitored on hosted
- PERFORMANCE ENGINEERING: Top ten performance issues on hosted and how they are mitigated
- ANECDOTES: Hosted war stories
will edit and add more if they come to mind.
Yes, would love a workshop like this.
Would echo all the topic suggestions given by @cryptovestor and @Koen.
Hey @cryptovestor , that sounds awesome. However, I think an end to end guide on getting setup as an indexer is something else we need.
Agreed - would you create a post making the request then we can flesh out what it would look like?
Sure @cryptovestor, will do.
Would like to see this also, adding the topic to the already brought up ones;
PERFORMANCE: comparing bare metal and docker/kubernetes setups, when to choose what
I would love a better understanding of the gateway selection if possible, especially stuff like this to be explained how and why it happened.
Maybe not related to performance, but a session on monitoring and alerts would be handy.
Could be interesting indeed, but yeah, i’d suggest making that different workshop. Feel free to make a fresh thread proposing this.
Adding to @Koen’s msg, @DAppsDeFiCrypto it would be great that when you create a new thread for this proposal to include details or questions (maybe even a detail of what you already have in place?) for context to kickstart the conversation
done
Any update on this workshop ? I’d love to gain understanding of performance & tuning. @martintel
Indexers expressed a lot of interest for this workshop in this week’s Office Hours