Resumable Online Index Creation In Azure SQL Database

Niko Neugebauer looks at a feature coming in SQL Server vNext:

It is about the time to create our first Clustered Online Resumable Index:

CREATE CLUSTERED INDEX CI_SampleDataTable	ON dbo.SampleDataTable (c1)	WITH ( ONLINE = ON, RESUMABLE = ON ) ;

But all we shall get is an error message:

Msg 155, Level 15, State 1, Line 25
'RESUMABLE' is not a recognized CREATE CLUSTERED INDEX option.

I was shocked and I was disappointed, but I have understood that it was my own mind’s fault. Nobody, I repeat – NOBODY has told me that it would work for the CLUSTERED Indexes, but when I see an announcement that the Indexes are supported, I was totally believing that the traditional (not XML, no CLR, no LOB’s) Rowstore Indexes would be totally supported. Oh yes, I know that it is crazy difficult. I know that this is a pretty forward-facing feature, but come on – my mind played trick on me, telling me the story that does not exist, for now, at least.

After realising my mind’s mistake I took a deeper breath and decided to try out the Resumable Nonclustered Index Creation with the following command:

CREATE NONCLUSTERED INDEX NCI_SampleDataTable	ON dbo.SampleDataTable (c1)	WITH ( ONLINE = ON, RESUMABLE = ON );

Hopefully we get a bit more support as SQL Server vNext is developed and eventually released.  In the meantime, Niko hits some limitations but his timings for the feature look good.

Related Posts

The (Un)Importance of Index Fragmentation

Tibor Karaszi argues that index fragmentation is a less serious issue than most DBAs think: You know the story. Every week or so, we defragment the indexes. Many of us uses Ola Hallengren’s great script for this, some uses Maintenance Plans, and there are of course other alternatives as well. But are we just wasting […]

Read More

The Performance Hit From Ignoring Duplicate Keys

Paul White explains why there is a big performance hit when using IGNORE_DUP_KEY on clustered indexes: The IGNORE_DUP_KEY index option can be specified for both clustered and nonclustered unique indexes. Using it on a clustered index can result in much poorer performance than for a nonclustered unique index. The size of the performance difference depends on how […]

Read More

Categories

July 2018
MTWTFSS
« Jun Aug »
 1
2345678
9101112131415
16171819202122
23242526272829
3031