Indexing vs. Sorting

Accepted Solution Solved
Reply
Occasional Contributor
Posts: 13
Accepted Solution

Indexing vs. Sorting

Hi,

 

I would like to add two simple indexes to a large dataset based on "column1" and "column2". Is adding the two simple indexes effectively the same thing as sorting the dataset on "column1" and adding an index based on "column2" assuming that the datasets will not be sorted again in the future?

 

The options I'm considering are:

 

proc datasets library=mylib;
modify largeDataset;
index create column1;
index create column2;
quit;

vs.

proc sort data=mylib.largeDataset;
by column1;
run;

proc datasets library=mylib;
modify largeDataset;
index create column2;
quit;

Wouldn't the second option be more space efficient than the first?

 

Thanks for your help!

Adam

 

SAS version 9.3


Accepted Solutions
Solution
‎10-31-2016 01:20 PM
Super User
Posts: 5,388

Re: Indexing vs. Sorting

No, it's not the same. The main feature of indexes is to allow for direct access to individual records (which in reality mens data page), which is beneficial for small subset operations.
Yes, by omitting an index you are saving space, but again, you will not get the same functionality.
Data never sleeps

View solution in original post


All Replies
Solution
‎10-31-2016 01:20 PM
Super User
Posts: 5,388

Re: Indexing vs. Sorting

No, it's not the same. The main feature of indexes is to allow for direct access to individual records (which in reality mens data page), which is beneficial for small subset operations.
Yes, by omitting an index you are saving space, but again, you will not get the same functionality.
Data never sleeps
Super User
Posts: 5,367

Re: Indexing vs. Sorting

There are other considerations besides storage space to hold an index.  Mainly, they focus on whether you are trying to retrieve all the observations vs. a subset.  Here's an earlier discussion that seems relevant:

 

https://communities.sas.com/t5/SAS-Procedures/Performance-SQL-vs-MERGE/td-p/117358

 

 

☑ This topic is solved.

Need further help from the community? Please ask a new question.

Discussion stats
  • 2 replies
  • 272 views
  • 2 likes
  • 3 in conversation