What is columnstore index and how is different from clustered and non-clustered

Question

Hi I am confused about columnstore index, what is column store index and how it is different from clustered and non-clustered index..

Answer 1

Columnstore index is very well explained here: http://www.patrickkeisler.com/2014/04/what-is-non-clustered-columnstore-index.html

The traditional clustered and non-clustered index you mentioned are both rowstore index, where the database stores the index row by row. The index will spread over several partitions, so even when we select only one column, the database still have to scan over all partitions to get the data, hence making a lot of I/O's.

Columnstore index, on the other hand, stores the index column by column. Normally, this will have all data of a column stored within one partition since all the data of one column combined is not that large. Now, when we select 1 column from the index, the database can return the data from one partition, which reduces a lot of I/O's. Moreover, columnstore index often has a significant compression ratio, therefore the I/O is even more efficient and the whole index can be saved in memory, which helps make queries 10x to 100x faster.

Columnstore index does not always perform better than rowstore. Columnstore index is suitable for scenarios like data warehousing and BI, where the data are often processed in mass, such as for aggregations. However, it performs worse than rowstore index in scenarios where data are often searched individual rows.

One thing worth noticing is that non-clustered columnstore index locks your table from being changed (but there are some work-around solutions to change the data), while clustered columnstore index still allows you to edit the data without dropping or disabling the index.

For more information on this topic, please refer to the article above and also try reading the MSDN documents.