The most effective combination is query caching plus an optimized semantic model . Oracle Analytics query caching stores previously generated query results so that compatible subsequent requests can be satisfied without repeatedly executing equivalent SQL against the underlying database. This reduces database workload, network traffic, and analytical response time for frequently repeated requests.

Caching alone, however, is not a substitute for sound semantic-model design. Proper logical relationships, aggregation behavior, source mappings, calculations, and physical connection configurations influence the efficiency of generated queries. Optimizing the model therefore addresses the cost of queries that do reach the source, while caching prevents unnecessary repetition of previously resolved requests.

Oracle also emphasizes that administrators need an appropriate cache-management strategy. Cache entries must be invalidated when source data or semantic-model definitions change so that users don't receive stale results.

Disabling caching increases database load. Duplicate subject areas add administrative complexity without inherently improving performance. Running every analysis every minute can substantially increase processing rather than reduce it.

Therefore, C provides the correct performance architecture: optimize query generation and use caching to avoid unnecessary source executions.

Reference Topic: Managing Administration, Security, and Performance — Query Cache and Semantic Model Performance

===============