Scale in testing.
WHY IT MATTERS
Behaviour at a hundred records differs entirely from a million.
WHAT CHANGES
Query performance Memory use Interface responsiveness Batch job duration Export and report generation
WHAT SMALL DATASETS CONCEAL
Missing indexes Queries scanning everything Pagination defects Timeouts
WHY MISSING INDEXES SPECIFICALLY
They are invisible until volume makes them fatal.
WHAT TO TEST AT VOLUME
Every list and search Every report Every batch process Every export
WHAT TO GENERATE
Data resembling production in distribution, not only in count.
WHY DISTRIBUTION MATTERS
Uniformly distributed generated data behaves differently from real data with concentrations.
WHAT TO REPRODUCE
Customers with far more records than average Text at maximum length Unusual character sets Old records alongside recent ones
WHAT TO MEASURE
Duration of operations, against the volume.
WHAT TO PROJECT
Behaviour at future volume.
WHAT TO ALERT ON IN PRODUCTION
Operations whose duration is growing.
