spike: product research on RPC + archive node offerings on other chains #530
Description
What problem does your feature solve?
There have been several discussion and spikes as of late that have delved into how we may implement the historical RPC features:
- https://github.com/orgs/stellar/discussions/1717
- https://github.com/orgs/stellar/discussions/1794
- Spike: Full history for
getTransaction(hash)#492
The decision points are often coming down to trade-off decisions centered around how much operator pain is tolerable. We don't have a great sense of exactly how much that is, which is hampering us in making the choice which optimizes the balance between that and consumer UX.
There have also been a number of feature requests that have recently come to light, and we'd like to have a better understanding of whether what's being asked is considered industry-standard, or can/should be delegated to an indexer or another downstream tool.
What would you like to see?
An analysis document that summarizes the takeaways from investigating and playing around with RPCs in other ecosystems. We want to be looking at things like:
- Storage size
- Storage mechanism (self-contained dbs/nodes vs external storage)
- Node/data initialization mechanism (ex. snapshot)
- Node/data initialization time
- Indexing/APIs available (notably, what the feature gaps are between these and our RPC)