Conditional data routing based on call type.

[TheFeloniousMonk]

In talks with @easy2stake about node efficiency, Alex has mentioned several times the need for smarter routing around request type. There's no need to run a full archival node to serve getBlock requests. Three archival nodes become one archival node, one full node, and one snapshot. In setting up infra this way, the cost for noderunners is reduced, and their margin increases.

This creates the need for conditional data routing, which will require the collaboration of both the protocol and the access point SDK (PATH or others). My first though is filtering PATH requests using the new on-chain metadata feature, with each filterable service being registered as the node type, for example:

• Eth Archival
• Eth Full
• Eth Snapshot

This would be especially helpful for heavy sources that usually only store limited data, like Solana (currently set for 6 epochs). Instead of generating a ton of errors or null requests, a gateway level response can indicate that full archival data is not available.

This is totally an open brainstorm on the best approach to this sort of function. The benefits of the outcome are clear from an infra perspective, but may have other unintended side effects (service bloat, etc.). We need to run through the options and decide on an approach.