---
title: "Announcing On-Demand Compute: Instant compute for your most intensive workloads"
description: "ClickHouse Cloud’s On-Demand Compute is announced in private preview as a way to run eligible queries on a shared pool of workers outside a service’s primary cluster, avoiding production contention and reactive autoscaling delays. It uses ClickHouse’s multistage distributed query execution and cost-based optimizer to distribute complex workloads, with queries requesting worker counts through settings including make_distributed_plan and distributed_plan_workers_num. Workers are leased for at least 60 seconds, reused during the lease, and released when it expires; concurrent queries share the pool, which grows to the largest request. In a stated benchmark, On-Demand Compute was faster for the tested workload and at 25 concurrent queries delivered similar performance while using 50% less compute, though results vary by query shape."
---

# Announcing On-Demand Compute: Instant compute for your most intensive workloads

[Clickhouse](https://yomu.fyi/company/clickhouse) · Sep 10, 2026

**Type:** Announcement

## Summary

ClickHouse Cloud’s On-Demand Compute is announced in private preview as a way to run eligible queries on a shared pool of workers outside a service’s primary cluster, avoiding production contention and reactive autoscaling delays. It uses ClickHouse’s multistage distributed query execution and cost-based optimizer to distribute complex workloads, with queries requesting worker counts through settings including make\_distributed\_plan and distributed\_plan\_workers\_num. Workers are leased for at least 60 seconds, reused during the lease, and released when it expires; concurrent queries share the pool, which grows to the largest request. In a stated benchmark, On-Demand Compute was faster for the tested workload and at 25 concurrent queries delivered similar performance while using 50% less compute, though results vary by query shape.

## Context

Metric-based autoscaling reacts to observed demand, so compute-intensive queries may wait for capacity. The post describes overprovisioning, waiting for autoscaling, and allowing heavy queries to compete with critical workloads as imperfect alternatives, motivating query-level scaling.

## Approach / What changed

Eligible queries request additional workers from a ClickHouse-managed pool using settings such as make\_distributed\_plan and distributed\_plan\_workers\_num. The capability uses multistage distributed query execution and a cost-based optimizer; workers are leased, reused during active leases, shared among concurrent queries, and released after lease expiry.

## Takeaways

- On-Demand Compute supports SharedMergeTree, Apache Iceberg, and Delta Lake workloads, including ad hoc queries, selected read offloading, and eligible data lake queries.
- Workers have a minimum 60-second lease that renews for longer-running queries; subsequent queries can reuse active workers without a cold start or discovery delay.
- In the benchmark, 25 concurrent queries using 15 workers achieved performance similar to the stateful cluster’s single-node execution while using 50% less compute; results vary by workload.

**Tags:** [ClickHouse](https://yomu.fyi/topic/clickhouse), [Cloud](https://yomu.fyi/topic/cloud), [Distributed Systems](https://yomu.fyi/topic/distributed-systems)

- Source: [Clickhouse](https://clickhouse.com/blog/on-demand-compute)
- Source URL: https://clickhouse.com/blog/on-demand-compute
- Ingested by Yomu: 2026-09-10T14:58:05.543Z

[Read original post](https://clickhouse.com/blog/on-demand-compute)