Query Scheduling

Kloudfuse runs many concurrent queries against a shared Apache Pinot cluster — interactive dashboard traffic, programmatic machine traffic, and scheduled background work all land on the same servers. Query scheduling decides how that capacity is shared so that small interactive queries are not blocked by large background ones, and so that high-priority workloads get served first when the cluster is busy.

Two mechanisms work together: query interleaving changes how a server distributes worker capacity across the queries that are currently running, and query priorities let operators set the relative importance and per-call timeout of each query.