Cloud Run serving a Firestore (Datastore mode) product list returns all products, causing memory limits and spikes in Datastore reads. To stop crashes and reduce reads while optimizing performance, what should you do?

Using cursors is the most effective solution because they allow you to paginate through large result sets efficiently without re-scanning previous results. This significantly reduces Datastore reads and memory consumption on Cloud Run by fetching only a subset of data at a time. Integer offsets are inefficient for large datasets as Datastore still has to scan through all preceding entities to reach the offset, leading to high read costs and slow performance. While limits reduce the number of items returned, they don't solve the problem of fetching subsequent pages efficiently without cursors, as each subsequent page request would still need to re-scan from the beginning if not combined with cursors. Increasing Cloud Run memory limits only postpones the problem and doesn't address the underlying inefficiency of the query or the high Datastore read costs.