Throughput refers to the volume of work completed per unit of time. For example, the amount of work performed by:
The number of a single transactions (for example, 100 transactions per second inserting trades from Wall Street).
All transactions across the server (for example, 10,000 read transactions per second plus 1,500 write transactions per second).
The number of reads performed (for example, the number of specific queries or reports per hour).
However, when you tune Adaptive Server for improved response times, you may decrease throughput, and vice versa. For example:
Adding indexes improves performance for queries and updates and deletes that use those indexes to avoid more expensive scans. However, you must maintain indexes during data manipulation language (DML) operations, which can decrease performance.
Using synchronous disk writes improves response times in a single transaction that includes a single user, but synchronous disk writes degrade multiuser throughput.