Design for concurrency

May include but is not limited to: hints, transaction isolation level, effect of database option READ_COMMITTED_SNAPSHOT, rowversion and timestamp datatypes

Most of this objective is covered in the previous post there’s simply too much overlapping in this part of the exam, as far as I can grab it. Anyway, there are some other interesting issues with locking and concurrency, so here’s a brief post on the remainders.

There are two data types related to concurrency – rowversion and timestamp. Rowversion is a binary data type which stores eight bytes, and increments automatically on every insert and update. Note that each table can have only one rowversion column, and you must give a name for it (you don’t have to specify a name for a timestamp value). Rowversion type is useful when you expecting multiple updates in roughly the same time, but I’ve stumbled into it when used LINQ to SQL along with WCF, too.

Continue reading “Design for concurrency”