Capture a reproducible operation
Record the database, user, document or report, start time, actual duration and target duration. For intermittent issues, correlate scheduled jobs, exchanges, backups and period-close activity.
Inspect every layer
Check client and network latency, 1C cluster processes, database waits and locks, CPU per core, memory pressure, storage latency and expensive application-code paths. A busy processor is evidence only when it aligns with the affected operation.
Collect the technology log to answer a question
Unfiltered logging creates large volumes and may distort the system. Use a short window and selected events to test a hypothesis: a lock, slow SQL statement, excessive server calls or memory growth. Keep lightweight operational monitoring permanently and enable detailed traces only for investigation.
Measure the business operation
Define targets for document posting, report opening or month-end closing. APDEX can aggregate satisfaction across operations, but each operation needs its own threshold. Validate the fix by repeating the same workload and checking that delay was removed rather than moved elsewhere.