Algorithms For Dummies

  Arranging and Searching Data 

     135

Of course, simply sorting the data isn’t enough. If you have an employee database 

sorted by last name, yet need to look up an employee by birth date, the sorting 

isn’t useful. (Say you want to find all of the employees who have a birthday on a 

certain day.) To find the birth date you need, you must still search the entire data-

set one item at a time. Consequently, sorting must focus on a particular need. Yes, 

you needed the employee database sorted by department at one point and by last 

name at another time, but now you need it sorted by birth date in order to use the 

dataset effectively.

The need to maintain several sorted orders for the same data is the reason that 

developers created indexes. Sorting a small index is faster than sorting the entire 

dataset. The index maintains a specific data order and points to the full dataset so 

that you can find what you need extremely fast. By maintaining an index for each 

sort requirement, you can effectively cut data access time and allow several people 

to access the data at the same time in the order in which they need to access it. The 

“Relying on Hashing” section, later in this chapter, gives you an idea of how 

indexing works and why you really need it in some cases, despite the additional 

time and resources needed to maintain the indexes.

Many ways are available to categorize sorting algorithms. One of these ways is the 

speed of the sort. When considering how effective a particular sort algorithm is at 

arranging the data, timing benchmarks typically look at two factors: