cf #230, bob's notes follow:
- Performance will be fine if you're fetching 10, 20, even 100 batches at a time. Databases are fast, and our aggregates only include the data that we need to do our calculations.
- if you honestly need to have more than 50 or so rows in an aggregate, lazy collections are your friends.
- In practice, we expect to have 20 or so batches "active" at any one time, and the rest are "finished". We only care about active batches for the purposes of this system.
- If your aggregates are loading too much data, and performance is suffering, you need to find another aggregate. There is no single "correct" aggregate, because it's a choice that we make in order to select an appropriate consistency boundary.
cf #230, bob's notes follow: