Software contention is the effect of multiple transactions competing for the same resources. Capacity Planner supports modeling software contention for software products that are affected by it. For example, when a write transaction starts, it locks its list or table and the transaction must complete before other transactions can access the same list or table. The effect is that other transactions must wait for the first transaction to unlock its list or table, causing transactions to queue. For capacity models that utilize locks, you can view their utilization on the Simulation Results Latency by transaction and Latency by site pages. An indirect effect that occurs when transactions enter a queue is that they have increased latencies, which you can also view on Simulation Results pages.

Methods to resolve highly-utilized locks and high transaction latency can be as simple as replacing storage devices or CPUs with faster ones or with those that have additional capacity. However, if this does not adequately improve latency, you might need to go as far as adding additional database roles to multiple servers and distributing the load equitability among them. Another way to decrease latency is by distributing usage among other devices.

See Also