ColumnStore Indexes UNIQUE and NOT DULL

Transcription

1

2 Agenda ColumnStore Indexes About me The Basics Key Characteristics DEMO SQL Server 2014 ColumnStore indexes DEMO Best Practices Data Types Restrictions SQL Server ColumnStore indexes

3 Gareth Swanepoel From South Africa originally Been in IT for 25+ years Was first a sysadmin, then a DBA Premier Field Engineer with Microsoft Author - SQL 2012 Pro Admin Speaker SQL User Groups, SQL Saturdays, Webinars, SQL PASS Summit Me@MyGareth.com Twitter Website :

4 ColumnStore Indexes the basics Row Store (Heap / B-Tree) Column Store (values compressed) ProductID OrderDate SaleAmount ProductID OrderDate SaleAmount data page data page data page data page

5 Key Characteristics of CSI's Columnar data format Key columns Faster query results Partitioning Record size limit Query processing Table cannot be updated WHAT????

6 DEMO SQL 2012 ColumnStore Index

7 PDW & SQL 2014 Column Segment C1 C2 C3 C4 C5 C6 Row Group:

8 ColumnStore Indexes PDW & SQL 2014 C1 C2 C3 C4 C5 C6 Delta (row) store C1 C2 C3 C4 C5 C6 Updates Column Store

9 PDW v2 & SQL 2014 xvelocity - Memory Optimised, Updateable CLUSTERED CSI ColumnStore is the preferred storage engine in PDW 2012 No secondary indexes are supported with CCSI s Tables are divided up into Row Groups & Column Segments Row Groups are approx. 1 million rows in size Index creation runs in parallel

10 PDW v2 & SQL 2014 DUI support is achieved by the use of a Delta table The Delta table is in row store format (no CCI) Delta rows are included into the CCI automatically (at approx. 1 million rows in size) CLOSED Delta rowgroups can be manually added with: ALTER INDEX <idx_name> REORGANIZE Deleted space can be reclaimed with: ALTER INDEX <idx_name> REBUILD PARTITION = <partition number all>

11 DEMO SQL 2014 ColumnStore Index

12 2014 ColumnStore indexes INSERTED a single record into a table with clustered columnstore index Screenshots taken from DMV sys.column_store_row_groups (subset of total rows returned) Before row inserted: After single row inserted (deltastore has been created and single row represented) After REBUILD REORGANIZE only moved delta store records into segment REBUILD affects entire index (or entire partition index) Segment row count increased by 1

13 2014 ColumnStore indexes The state_description field has three states: COMPRESSED, OPEN, and CLOSED COMPRESSED represents a row group that is stored in columnstore format OPEN represents a deltastore that is accepting new rows CLOSED represents a full deltastore ready for REORGANIZE When inserting 102,400 rows or more in a single batch into a columnstore index distribution, the data will compress automatically When inserting 102,399 or less rows in a single batch, the data will be stored in the delta store The actual maximum number of rows per delta store is 1,048,576 at which point it is CLOSED This is also the ideal segment size that SQL Server will try to create when first building a columnstore index from a table When Index Build encounters memory pressure, DOP is reduced first and then segment size is reduced Only the REBUILD statement can compress a delta store that is not in the CLOSED state Neither REORGANIZE or the Tuple Mover process will have any effect on an OPEN delta store

14 2014 ColumnStore indexes Consider a table with a single partition range of 100 with two partitions INSERT the partition key values 50 and 150 into the table REBUILD the index on partition 1 INSERT three more records into partition 1 (Note partition 2 is still in deltastore form also) REBUILD the entire index (as opposed to the single partition above) Note: When attempting a partition MERGE or SPLIT operation, be aware that a table with a columnstore index will return an error and fail to execute whereas a traditional row-based clustered index/heap will perform the operation successfully

15 Best Practices Choosing Columns (Non-Clustered CSI only) DO put CSI's on large tables only DO structure queries as star-joins DON'T use OUTER JOIN DON'T use UNION ALL DON'T use string filters, etc DON'T use NOT IN (<subquery>)

16 Supported Data Types char and varchar nchar and nvarchar (except varchar(max) and nvarchar(max)) decimal (and numeric) int, bigint, smallint, tinyint and bit float (and real) Uniqueidentifier (2014 only) money and smallmoney All date and time data types binary and varbinary (except varbinary(max))

17 Unsupported Data Types ntext, text, and image varchar(max) and nvarchar(max) varbinary(max) rowversion (and timestamp) sql_variant CLR types (hierarchyid and spatial types) Xml

18 Restrictions Cannot have more than 1024 columns (SQL limitation) Cannot be a clustered index (2012 only) Cannot be a unique index (with caveats) Cannot act as a primary key or a foreign key (with caveats) Cannot be created on a view or indexed view Cannot include a sparse column Cannot be changed using the ALTER INDEX statement Cannot be created by with the INCLUDE keyword Cannot include the ASC or DESC keywords for sorting the index

19 Feature Restrictions Page and row compression, and vardecimal storage format Replication Change tracking Change data capture FileStream FileTable

20 ColumnStore Indexes in SQL & Azure SQL DB, DW, APS Snapshot isolation and read-committed snapshot isolation Specify ColumnStore index when creating a table Updateable non-clustered ColumnStore indexes AlwaysOn readable secondary supports updateable ColumnStore indexes ColumnStore index on a memory-optimized table Filtered non-clustered ColumnStore indexes Performance enhancements 2017 only non-persisted computed columns

21 References Books On-Line: ColumnStore Indexes Versioned Feature Summary: Niko Neugebauer ColumnStore Blog Series:

22 Thanks Me@MyGareth.com Twitter Website :