Summary

SQL Server lets you manage multiple users simultaneously and ensure that transactions observe the properties of the chosen isolation level. Locking guards data and the internal resources that make it possible for a multiple-user system to operate like a single-user system. You can choose to have your databases and applications use either optimistic or pessimistic concurrency control. With pessimistic concurrency, the locks acquired by data modification operations will block users trying to retrieve data. With optimistic concurrency, the locks will be ignored and older committed versions of the data will be read instead. In this chapter, we looked at the locking mechanisms in SQL Server, including full locking for data and leaf-level index pages and lightweight locking mechanisms for internally used resources. We also looked at the details of how optimistic concurrency avoids blocking on locks and still has access to data.

It is important to understand the issues of lock compatibility and escalation if you want to design and implement high-concurrency applications. You also need to understand the costs and benefits of the available concurrency models.