Data-access audit
When latency spikes correlate with concurrent users rather than single heavy queries, the issue often lives in how services open connections and scope transactions.
We trace representative HTTP or message-handler flows into repository calls, counting round trips per request and flagging patterns that amplify load: chatty loops, implicit lazy loads, long-lived transactions, and mis-sized pools.
Deliverables
- Sequence diagrams for three to five critical paths you nominate
- Table of round trips per path with suggested consolidation or batching
- Connection pool and timeout settings reviewed against peak concurrency captures
- Read replica routing checklist if you use replicas
Duration
Three to four business days after access is granted. Starting from R 32 000—see rates.