When we opened Watasu in April, ClickHouse shipped in early access — and quickly became the add-on you asked about most. Today it’s generally available, with a full plan ladder from a €15 dev instance to multi-shard clusters built for billions of rows.
What ClickHouse is, and why you might want it
Most apps keep their data in a transactional database like PostgreSQL — and they should. It’s built for many small reads and writes: load this user, update that order. But ask it a different kind of question — “average session length per day, per country, over the last six months” — across a hundred million rows, and you’ll feel it. Row-oriented databases store data record by record, so wide analytical scans churn through everything.
ClickHouse stores data by column instead. When a query touches three columns out of forty, it reads exactly those three — compressed, in big sequential chunks, in parallel. The result feels like a different category of tool: aggregations over millions or billions of rows that return in seconds, on append-heavy data that would make a transactional database weep.
Good fits, from what we’ve seen so far:
- Product analytics — every click, page view, and feature event, queryable while it’s still being written
- Audit logs at scale — append forever, slice instantly when someone asks “who changed this?”
- Time series — metrics, sensor data, anything with a timestamp and a firehose behind it
- Dashboards — wide aggregations that stay interactive instead of timing out
And the honest counterpart: don’t use it as your transactional database. ClickHouse is not built for OLTP, and you’ll hate it for that. Keep PostgreSQL for the orders; let ClickHouse explain them.
One command, like everything else
watasu addons:create clickhouse --app my-appConnection details land in your app’s config vars; any standard ClickHouse client library works. Need to poke around interactively?
watasu clickhouse:cli --app my-appA plan ladder with an honest gap in it
Three tiers, and the jump between the first two is deliberately more than “a bigger box”:
- Hobby (€15–€60/mo) — a single node on network-backed storage. Explicitly non-production: it’s for development, exploration, and small staging, with manual backups only.
- Standard (€80–€260/mo) — production begins here, and the topology changes underneath you: one shard with two replicas across two German zones, on local NVMe, with scheduled backups by default.
- Premium (€600–€2,200/mo) — multi-shard scale-out with two replicas per shard and distributed tables, for datasets that outgrow a single node.
We’d rather draw that line clearly than let a “small production” instance quietly be neither. The full rate card is on the pricing page.
Backups that assume the worst
Like every Watasu database, ClickHouse restores are non-destructive: a restore creates a replacement add-on you can inspect, and you promote it only when you’re satisfied. Capture a manual backup any time with watasu addons:backups:capture; Standard and Premium add scheduled backups on top.
One tip from the team: validate your restore path before you need it. ClickHouse archives can be large, and an incident is the wrong moment for surprises. While you’re at it — ClickHouse loves materialized views, so pre-aggregate the queries your dashboards run most.
Get started
Attach one to a staging app and point your event stream at it — the ClickHouse docs cover the details, and the pricing page has every plan. Questions, or a workload you’re not sure fits? We’re at info@watasu.io.
Happy querying — and may the fourth be with your aggregations.