PostgreSQL consulting
Your PostgreSQL, fixed by people who run it.
Slow queries, memory alerts, a cloud bill that keeps growing, an upgrade nobody wants to start. We find the cause and fix it, usually without a bigger server.
01 Results
Fixes you can
measure.
-
For a customer
$630 → $320
a month
Memory alerts at 93% on a multi-tenant server with 82,000 tables. One connection-pooler setting, half the memory, and the alerts stopped.
Read how we did it -
In our own production
150 GB → 0
of disk spill a day
An order API fell to half its volume in festival season, with no deployment to blame. The fix was one line of SQL, scoped to one function.
Read how we did it -
In a load test
½ the CPU
same query, same result
An hourly job called one SQL function 10 million times. Rewriting that function in PL/pgSQL, without touching its query, halved the CPU.
Read how we did it -
For a stock-market platform
2,000 → 10,000
transactions a second
Same hardware, no new instances. A rethink, end to end: the application, the database, and the load between them.
02 What we take on
The PostgreSQL work
that keeps you up.
-
Slow queries and locks
Plans, indexes, lock chains, vacuum and bloat, sorts spilling to disk: found from the database’s own statistics, fixed where the cause is.
-
Memory and the cloud bill
Right-sizing instances, connection pooling with PgBouncer, and the settings that decide whether you need a bigger server. Usually you do not.
-
Upgrades and migrations
Major-version upgrades, moves between clouds and managed services, and migrations from other databases: planned, rehearsed, then landed.
-
Replication, backup and failover
Replicas that fall behind, slots that hold WAL, backups and point-in-time recovery you have actually restored from.
-
PL/pgSQL and schema design
Functions, triggers and data models reviewed by the co-author of a PL/pgSQL book, for speed and for the next person who reads them.
-
When production is down
Rescue on a bad day: we join the call, find the cause with you, and write down what happened so it does not happen again.
03 How it works
Small first step,
no surprises.
-
01
A free 30-minute call
Tell us what is slow, expensive or worrying. We say what we would look at first, and whether we are the right people for it.
-
02
A read-only look
We start without changing anything. A role with pg_monitor is usually enough to find the cause, and you see everything we run.
-
03
The fix, measured
We write up the cause and the fix, apply it with your team, and measure before and after so the result is a number, not a feeling.
04 Who you get
Fifteen years of
running PostgreSQL.
Managed PostgreSQL platforms built from scratch, monitoring watched in production, migrations landed across versions and clouds. The same people answer your call.
AWS Oracle EnterpriseDB OpenSCG Tessell
Dinesh Kumar
Founder
Co-author of Procedural Programming with PostgreSQL PL/pgSQL and the PostgreSQL High Performance Cookbook. Builder of pgSherpa.
Tell us what is slow.
Thirty minutes, free, with the person who would do the work. If we are not the right fit, we will say so.