The docs indicate[1] that electric's syncing logic, "shapes", are handled transactionally. How would electric handle an initial sync of 500mb or an initial sync of 10gb?
Apps with large dataset have their own characteristics: they are more tolerant to slower initial sync. We can resume replication at point-in-time, so even if a shape takes long to load, we can resume replication at the point we started synching the shape.
We will be working on the engineering to handle transferring large shapes.
