CHAPTeR 9 DXP PeRfORMANCe OPTIMIzATION • A caching framework was used to cache the master lookup values (overall list of suppliers, list of countries, etc.). • LDAP objects were cached to avoid multiple LDAP queries. Database Layer Performance Optimization Following are the performance optimizations done at the database layer: • Database indexes were created for the columns appearing in the filter conditions of the queries. The table data was partitioned based on the country, as suppliers and distributor data were mainly using country-specific values. • Denormalized tables (a method in which table structure is flattened out with all needed columns to avoid real -time complex joins) were created to hold the results of complex table joins. These snapshot tables were refreshed every 5 hours. This avoided costly real-time table joins, improving the query performance. • All the database-specific configurations (such as checkpoint logs, query cache size, sessions, log buffer, recomputation of statistics, etc.) were optimized from a performance standpoint. • Lookaside database tables and materialized views were created to store the results of frequently used queries. Chapter Summary • The main performance best practices for presentation layer performance optimization are: reducing HTTP requests, image optimizations, optimal resource loading, caching, delivery from edge locations, compression, and optimal positioning. • Key performance metrics are: TTFB, page size, page response time, above the fold time, and perceived response time. 258