Performance dreams of sql server 2014

Post on 05-Dec-2014

603 views 1 download

description

Performance dreams of sql server 2014 session at SQL Saturday Peru event on 7/12/2013

Transcript of Performance dreams of sql server 2014

04/10/23 by Shehap El Nagar

Shehap EL-Nagar

I am MVP,MCTS , MCITP SQL Server, I am DB consultant and Architect for lots of Banking, Telecom ,Ministries and governmental organizations all over Gulf ,also he has deep knowledge about T-SQL performance , HW Performance issues, Data Warehousing solutions , SQL Server Replication, Clustering solutions and Database Designs for different kinds of systems ...

The founder of the biggest SQL Server community all over the middle east http://sqlserver-performance-tuning.net/ , you can watch its success memories at http://www.youtube.com/user/ShehapElNagar

Moderator and author at http://www.sql-server-performance.com ,, the 1st SQL Server Author at MSDN Arabia http://msdn.microsoft.com/ar-sa/library/jj149119.aspx

, Speaker at SQL Saturday Events worldwide , local events at Saudi Arabia , many online events , more than 90 video tutorials and also many private sessions for .net developers and Database Administrators

And also influent participator at Microsoft Forums of SQL Server at http://social.technet.microsoft.com.More about him , you can find him on MVP Microsoft site   http://mvp.microsoft.com/en-us/mvp/Shehap%20El-Nagar-5000188 .You can contact him at the below contacts :Mail :idgdirector@yahoo.com ….Cellular phone :00966560700733  

First :Overhead look on new SQL Server 2014 features

• Definition • Architectural design • Values and benefits• Caveats and recommendations• Demo

Third: Single Partition OIR , locks priority management

• Major Improvements outlines • Evolution from SQL Server 2000 to SQL Server 2014

Second :In Memory OLTP built in

Third :CCI (Clustered Columnstore Index)• Definition • Architectural design • Values and benefits• Caveats and recommendations• Demo

• Definition • Values and benefits• Demo

Fifth: New HW Resources supported and IO Resource Governor

First :Overhead look on new SQL Server 2014 features

The Evolution of Microsoft Data Platform

SQL Server2000

SQL Server2005

SQL Server2008

SQL Server2008 R2

SQL Server2012

XML ● KPIs

Management Studio ● Mirroring

Compression ● Policy-Based Mgmt ● Programmability

PowerPivot ● SharePoint Integration ● Master Data Services

Always On ● ColumnStore Index ● Data Quality Services ● Power View ● Cloud Connectivity

SQL Server2014

In-Memory Across Workloads ● Performance & Scale ● Hybrid Cloud Optimized ● HDInsight ● Cloud BI

0 211 8

0 1 4 0

69

53 53 55

34 2922 18

5

28 25 2921

6 120

16 169 8 6

43

20

43

0

20

40

60

80

2006 2007 2008 2009 2010 2011 2012 2013SQL Server Oracle MySQL DB2

Database Vulnerabilities

Secure

*National Institute of Standards and Technology Comprehensive Vulnerability Database 4/17/2013, Market share from IDC 2013

In-Memory OLTP10X-30X faster for new and existing SQL Server apps

Clustered column store indexUpdatable clustered columnstore index

Single Partition OIRSingle partitions for online index rebuild for a less impact

Locks priority ManagementDefine priority level and wait times for OIR for a less impact

Hardware scalableIO Resource Governor and more HW resources are supported

First Myth: In Memory OLTP

Architectural design Hekaton project to reduce the cost of query engine (35% delay) and access engine (45% delay)

Memory optimized tables are working on memory as streamlined

No page latch, no buffer pool , no lock manager like disk based table

They are supporting Natively compiled Stored procedures which are .dll entry points compiled to machine code using Visual C++ compilers

SQL Server integration

• Same manageability, administration, and development experience

• Integrated queries and transactions

• Integrated HA and backup/restore

Main-memory optimized

• Indexes (hash and range) exist only in memory

• No buffer pool• The Highest

durability• Columns size

from 8060 bytes to 512 GB512 GB 

High concurrency

• Multiversion optimistic concurrency control with full ACID support

• Core engine uses lock-free algorithms

• No lock manager, latches, or spinlocks

T-SQL compiled to machine code

• T-SQL compiled to machine code via Visual C compiler

• Invoking a procedure is just a DLL entry-point

• Aggressive optimizations at compile time

• No change whatsoever in your codein your codeSteadily declining

memory price, NVRAM

Many-core processors

Stalling CPU clock rate

TCO

Hardware trends Business

Hybrid engine and integrated

experience

High-performance

data operations

Frictionless scale-up

Efficient, business-logic

processing

In Memory OLTP benefits and values

In Memory benefits and values

Microsoft In-Memory Technologies

• In-Memory Analytics in PowerPivot for Excel

• In-Memory Analytics in SQL Server Analysis Services

• StreamInsight • In-Memory OLTP

• In-Memory Columnstore Index (in SQL Server 2014)

2014

2012

2014

2012

2014

Caveats and Recommendations :•Identity and Sequence

•Foreign keys and constraints

•DML triggers

•Some columns types like BLOB columns as image and varbinar (ma), XML, ROWGuid, timestamp, text as well as ntext and CLR columns as hierarchyid,geography and geometry

•Data Compression and Clustered index but just only NONCLUSTERED HASH index

•Computed columns.

•Truncate option but just delete

•Table hints like with (nolock)..etc

•Alter natively compiled procedure is not supported but just you have to drop it and re-create it again•Natively compiled SPs don’t support merge commands

Demo Part

Second Myth: Clustered ColumnStore Index

• Nonclustered column store indexes are based on xVelocity analytics engine and xVelocity columnstore index

• Store data in column-wise fashion 

• Hekaton technique which is in contrast a

row-based technology 

• CCI (Clustered column store index ) is a mix between Hekaton and xVelocity analytics engine and xVelocity columnstore index

• CCI is the table itself as CCI includes all columns

Architectural design

• Columnstores: clustered and updatable No contradiction with DML as Data can be inserted, updated, or deleted

• Fast execution for data warehouse queries• Speedups of 10x -30x

• No need for separated base table• Saves space and Simpler management

• Eliminate need for other indexes• Saves space and simpler management

• More data types supported for batch mode execution• Removes many limitations from non-clustered columnstores in SQL

Server 2012

Benefits and values of CCI

Benefits and values of CCIBatch Mode Row Mode

Caveats and Recommendations :

•Foreign keys and constraints

•DML triggers

•Some columns types like BLOB columns as image and varbinar (max), XML, ROWGuid, timestamp, text as well as ntext and CLR columns as hierarchyid,geography and geometry

•Computed columns.

• Transactional replication nor peer to peer replication

Demo Part

Third Myth: Single Partition OIR and

locks priority management

Definitions and terminologies

Managed Lock Priority is used for determining the lock priority of OIR (online index redbud) or switch partitions relatively to other end users transaction

Single Partition Online Index Rebuild  is used to allow OIR for single partitions

Benefits and values

• Increased concurrency and application availability

• Saving for throughput and thus resources such as CPU , Memory  and IO

• Log space is reduced much coz not all tables are going to be rebuilt once a time but smaller chunks/partitions which permit for faster checkpoints

Fourth Myth: New HW Resources supported

and IO Resource Governor

New HW resources and IO Resource Governor

Demo Part

Q & APost your questions at:

http://www.sqlserver-performance-tuning.net/forums/

04/10/23

Thank you ..See you again