[Search] Searching across federated data

[pmn]

How should search work across federated / distributed data? Where do the indexes live? Is this a potential point of attack for actors who wish to block access to specific documents?

[daeken]

The search will not go beyond a single node, to keep things simple. I'm imagining search will be limited to a given set of documents (e.g. one leak) and only on one node, for simplicity and scalability's sake.

Is there any reason to search beyond one node?

[pmn]

I suppose it depends on whether or not you think split up documents should be distributed across multiple nodes for durability purposes.