SQL Server Fill Factor – Behavior of Data in Indexes

fill-factorindexsql serversql-server-2008-r2

Let's say you have a database where the default fill factor is 20. Whenever data is inserted, does it only create pages filled up to 20%?

From my understanding, when the data is inserted there will be roughly 20% of data in the pages. When the data is updated, however, it will expand to more than 20% of the index, upto filling it and generating a page split, right?

Best Answer

Fill factor only comes into play when an index is created or rebuilt. It is the amount of consumption for the index of the leaf level pages that are filled during these operations. (see the note below for more clarification on affected page levels)

When there is a DML command to data (INSERT, UPDATE, and/or DELETE), it will happen to the corresponding affected indexes. In other words, if you have a page that is 20% filled and you insert data into that page, the page will contain more than 20% of the data (let's say 35% just for example's sake). Do another insert, now the page is 64% filled. Rebuild the index, and the leaf level pages will now relatively contain the space percentage that you specify (or implicitly the default value for the server).

(Note, when you don't specify PAD_INDEX to be ON, fill factor is only applied to leaf level pages. But when you set PAD_INDEX = ON, the fill factor will be taken into account for intermediate-level pages of the index. The default is OFF)

The reason to adjust fill factor (instead of using the default 100/0) is so that you minimize page splits when inserting or updating data. But keep in mind, nothing is for free. The lower the fill factor, the more space data would normally take up. If you keep an 80% free page space for your indexes, they will consume a relatively larger amount of disk space, which can lead to more reads.

From my understanding, when the data is inserted there will be roughly 20% of data in the pages. When the data is updated, however, it will expand to more than 20% of the index, upto filling it and generating a page split, right?

When data is inserted it will insert into the appropriate indexes at the appropriate page. This could and most likely would very well cause the page consumption to be higher than the fill factor.

A page split will happen when new data is added to a full index page. SQL Server will then split the page and approximately place half of the data from the full page into a new page. Again, fill factor does not come into play here.

A legitimate reason to lower fill factor would be to minimize page splits, therefore minimizing index page fragmentation.

Related Solutions

Sql-server – PAD_INDEX and FILLFACTOR on clustered Identity index

PAD_INDEX only applies to non leave level pages. By default those are filled full.

So specifying PAD_INDEX=ON together with FILLFACTOR is only useful when you specify a fillfactor smaller then 100%.

In your case, specifying PAD_INDEX is not necessary, since by default your leave pages are how you want them. Full.

Sql-server – index maintenance strategy if few insertion points relative to # of rows

The answer by ThomasStringer is very good and the usage of partitioning here would likely help overall performance and decrease your maintenance cost - however, it won't do anything for your page split scenario.

Can you verify that your FK's are trusted? This won't decrease or impact your page splits but it will help your query execution time.

select 'ALTER TABLE ' + s.name + '.' + OBJECT_NAME(o.object_id) + ' WITH CHECK CHECK     CONSTRAINT ' + i.name + '
GO'
FROM sys.foreign_keys i  INNER JOIN sys.objects o ON i.parent_object_id = o.object_id     INNER JOIN sys.schemas s ON o.schema_id = s.schema_id WHERE     i.is_not_trusted = 1 AND i.is_not_for_replication = 0
-- Foreign Key Check... Will build the statement below to CHECK (validate) the FK for     those FK's that are untrusted.
UNION
select 'ALTER TABLE ' +  s.name + '.' + OBJECT_NAME(o.object_id) + ' WITH CHECK CHECK     CONSTRAINT ' + i.name + '
GO'
from sys.check_constraints i join sys.objects o on i.parent_object_id = o.object_id     join sys.schemas s on o.schema_id = s.schema_id where i.is_not_trusted =     1 and i.is_not_for_replication = 0

-- Check Constraint Check... Will build the statement below to CHECK (validate) the       check constraints for those check constraints that are untrusted.

-- !!!! -- The output (in TEXT) will look similar to what you see below.  once you     have the output, put it into the query window and execute it.

Another idea you already hit on, using fill factor - if you use a low fill factor and combine that with either row or page level compression, your data density, per page will increase. Again, this won't help your page split problem, BUT it can decrease the quantity of page splits (due to having a higher data density per page).

Lastly, if you can, look at changing your index and lead with the log_time column instead of the foreign key column. This change, depending on how the log_time data comes in (hopefully it's more "in order" than the FK column). This could decrease your page splits significantly. If you pair this with both compression (ROW) and a good partition scheme, you might see some significant improvements.

Best Answer

Related Solutions

Sql-server – PAD_INDEX and FILLFACTOR on clustered Identity index

Sql-server – index maintenance strategy if few insertion points relative to # of rows

Related Question