A predictable, transparent process.
Seven steps from "things are slow" to "shipped and verified" — every step visible inside your dashboard.
-
01
Discovery
15–30 minWe start with a 30-minute call (or async write-up). Tell us the workload, the engine, the pain — and what success looks like for you.
-
02
Access & sandbox
Day 1You provide read-only credentials or a sanitised schema dump. We build a staging copy that mirrors production shape but never holds production data.
-
03
Audit
Day 1–2We profile slow logs, run EXPLAIN ANALYZE, pull pgBadger / MySQLTuner reports, and rank fixes by expected impact-vs-risk.
-
04
Plan
Day 2–3You receive a written plan: every change, every benchmark target, every rollback. You sign off before any write hits a real cluster.
-
05
Stage & benchmark
Day 3–5Changes land on staging first. We run before/after benchmarks on production-shaped traffic and verify replication lag, locks, and row counts.
-
06
Cutover
Cutover windowFor migrations: a logical-replication or dual-write cutover, typically under 5 minutes. For tuning: rolling apply with a verified back-out at each step.
-
07
Verify & hand off
30-day windowPost-deploy, we monitor for the agreed window, deliver a PDF report, and leave the chat open for 30 days for follow-up questions.
Battle-tested tools, opinionated choices.
We use what works on production workloads — not what looks shiny in a demo.
Plan analysis on every problem query, with row estimates vs actuals.
Postgres slow-log aggregation and time-series analysis.
MySQL config audit + query digest for the slow log.
Lossless cross-engine migrations with type mapping.
When you need ongoing replication during a long cutover window.
Reproducible, validated backups — never trusted blindly.
Want to see the process on a real workload?
Start with a Quick Audit — you'll have a punch list within 24 hours.