The demands on the system keep on increasing, in terms of both log entry ingestion and servicing user queries. If anything's changed, then I think it's all load-related.
I'll see if our Ops team can provision some additional readonly replicas to help handle the higher load.