Implement the inbound edge lookup and storage interfaces in Cassandra

changed milestone to %Tooling for takedown notices [Roadmap - Collect]

added activity::Implementation label

marked this issue as related to #4680 (closed)

After some preliminary research, it seems we need to manually create a table for each week, and then manually construct table names for queries. The primary key will be the target object (as this is what we will need to match on).

Contrary to what it seems, we don’t care about TimeWindowCompactionStrategy because we are not interested in grouping data by time. We are just interested in knowing if new edges have been added from those we already know.

mentioned in merge request !1045 (merged)

mentioned in issue swh/meta#4803 (closed)

We've implemented the basic logic for Cassandra.

Before we are able to write object_references in cassandra, we need to extend the code to actually support time-based partitioning with pruning of old partitions, which needs a bit more design work in our own CQL layer.