Your ClickHouse bill creeps up every month — runaway queries scanning billions of rows, parts piling up, storage nobody set a TTL on. OptiHouse connects read-only, finds exactly what's burning compute and disk, and writes the optimized SQL for you. No agents, no code changes, no risk.
$pip install optihousetest it locally — no signup, no cluster Illustration of one scan — OptiHouse runs it for you inside the dashboard. No CLI, nothing to install.
Every Wave 1–7 rule fires instantly against your SQL — anti-patterns, PREWHERE, FINAL, JOIN algorithm, projections, skipping indexes. We never run your query — only analyse the text.
Install the CLI and scan your own ClickHouse from your machine — read-only, nothing leaves your network. See exactly where the money is before you ever sign up.
pip install optihouseoptihouse scan --demooptihouse scan --host db.internal --user readonly --password '***'pip install optihouse · MIT · prefer a UI? the same analysis runs here in the browser.
Every scan returns prioritized fixes with the exact SQL and estimated impact. Here's what the Recommendations screen looks like.
Illustrative example data — your results depend on your cluster.
What changes once OptiHouse ranks every fix by its dollar impact.
Leave your email and we'll send you the OptiHouse deck — how it finds wasted ClickHouse spend and how much it saves.
No spam. Just the deck — and answers to your questions if you want them.
Most ClickHouse clusters have 20–40% of spend tied up in waste. Try your number.
Estimate based on a typical 20–40% reclaimable range. Your result depends on your cluster.
Add a ClickHouse host with a read-only user. Optional SSH bastion for VPC-locked clusters.
We collect query_log, system.parts, system.replicas. No agents on your boxes — pull-only.
Each finding ships with exact SQL, expected impact in dollars and a confidence score.
From the first scan to the last optimization — every tool a data engineer needs in one place.
Catches SELECT *, missing WHERE, function-on-partition-key, repeated heavy aggregations — and more silent killers.
Per-table size, part fragmentation, tiny-parts and tables nobody has read in 90+ days — the exact GB you can reclaim today.
Finds poorly-compressed columns and hands you the exact ALTER … CODEC to shrink them 2–5×.
Compares every scan — spots queries that suddenly read 10× more bytes or run 5× longer than before.
Paste any query — get a before/after with the exact partition-friendly version, ready to copy.
Replicas, mutations, merges and disk capacity at a glance, with days-until-full forecasts on every disk.
Whether you're paged at 3am or you sign the cloud bill — OptiHouse surfaces the right signal for your role, not a wall of metrics.
Auto-generated runbooks for stuck mutations, merge backlog, replica imbalance. Every recommendation ships with diagnostic SQL — ready to paste.
Per-team chargeback for ClickHouse compute, weekly Slack digest with $-savings impact, upgrade-readiness risk before you bump versions.
Per-column codec audit, zombie MV detection, projection advisor, PK alignment vs real WHERE patterns. Schema intelligence that knows ClickHouse internals.
Drop your manifest.json → every recommendation deep-links back to the dbt model. The PR bot posts inline lint findings on every SQL change before it ships.
Cold-data-tier savings, a pre-flight cost estimator before queries run, plus a Prometheus + OpenTelemetry exporter so OptiHouse plugs into your existing dashboards.
We're a small team that ran ClickHouse in production for years and got tired of writing the same diagnostic SQL at 3am. OptiHouse is the tool we wished we had.
“Every ClickHouse cluster I'd ever run had the same hidden taxes — zombie MVs, FINAL on every read, partitioning that never pruned, codecs nobody picked. The tools at the time would tell you something was off; none of them told you what to do next. So we built one that does.”
Every plan includes the SQL Optimizer, the core advisors and read-only access. You upgrade only when you need more clusters or enterprise controls.
The full Team feature set, free for 14 days.
For a single team running one or two clusters.
For growing data orgs running multiple clusters.
Regulated, multi-cloud, audit-grade.
Sign up, connect your cluster (read-only), and we'll surface the top fixes ranked by impact — with the SQL to apply them.