Case 01 · E-commerce
An order database that collapsed on campaign day
Situation: On a national e-commerce platform, order listing and payment steps climbed to 3–8 seconds during sales periods, with frequent timeouts.
What we did: Extracted the most expensive queries, keyset pagination, targeted covering indexes, monthly partitioning on the hot table and moving report queries to a read replica.
5×faster p95 (2.9s → 0.34s)
0campaign-day timeouts
14 daysfrom first result to production
+12 mohardware upgrade deferred
Case 02 · Fintech
From a single server to multi-region high availability
Situation: A payments infrastructure provider ran on a single primary database and took minutes of downtime during maintenance and failures.
What we did: Always On Availability Groups, automatic failover, retry and idempotency on the application side, quarterly DR drills and a runbook.
99.99%measured availability over 12 months
< 40sRTO measured in drills
0downtime during planned maintenance
4successful DR drills per year
Case 03 · SaaS
From on-prem to a managed cloud database
Situation: A B2B SaaS product carried growing maintenance load and hardware risk on servers in its own data centre.
What we did: Compatibility analysis, landing-zone setup (IaC), a two-rehearsal cutover with online replication, post-migration right-sizing and a reservation plan.
8 mintotal cutover time
42%lower monthly infrastructure cost
0data loss / inconsistency
60%reduced operational workload
The figures on this page are representative examples from real projects and are labelled as such on the page. Client names and signed reference letters will be added after publication approval (see README.md).
