To maximize the benefit from nonclustered indexes, you must minimize the cost of the data retrieval as much as possible. A major overhead associated with nonclustered indexes is the cost of excessive lookups, formerly known as bookmark lookups , which are a mechanism to navigate from a nonclustered index row to the corresponding data row in the clustered index or the heap. Therefore, it makes sense to look at the cause of lookups and to evaluate how to avoid this cost.
In this chapter, I cover the following topics:
The purpose of lookups
The drawbacks ...
