Stretch database – Modifying the filter function

-> Please refer “https://jbswiki.com/2017/06/15/stretch-database-in-sql-server-2016/” if you want to setup Stretch database from scratch. If you refer the article I would have named the function “StretchByYear” in the wizard page “Select rowst  stretch”.

-> I will open the function StretchByYear to check what it contains,

Blog6_1.PNG

-> The modify script for the function is as below,

USE [JB_StretchDB]
GO
/****** Object: UserDefinedFunction [dbo].[StretchByYear] Script Date: 17/06/2017 7:59:56 PM ******/
SET ANSI_NULLS ON
GO
SET QUOTED_IDENTIFIER ON
GO
ALTER FUNCTION [dbo].[StretchByYear] (@Year Int)
RETURNS TABLE
WITH SCHEMABINDING
AS RETURN SELECT 1 AS is_eligible
WHERE @Year <= CONVERT(Int, N’1982′)

-> The modify function updates us that all rows less than or equal to 1982 will be moved to the cloud.

-> Checking the stretch monitor, We see 20000 rows on On-Premise and 30000 rows on cloud.

Blog6_2.PNG

-> The current filter function moves all data less than equal to 1982 to the cloud. Now lets consider that the  requirement changes to move all data less than equal to 1983 should be moved to cloud. In order to achieve this, I will create a new function as below,

USE [JB_StretchDB]
GO
CREATE FUNCTION [dbo].[StretchByYear_1983] (@Year Int)
RETURNS TABLE
WITH SCHEMABINDING
AS RETURN SELECT 1 AS is_eligible
WHERE @Year <= CONVERT(Int, N’1983′)
GO

-> One thing to remember is, the new filter function has to be less restrictive than the previous function.

-> Enabling the filter function for table Table1,

ALTER TABLE Table1 SET ( REMOTE_DATA_ARCHIVE = ON (
FILTER_PREDICATE = dbo.StretchByYear_1983(Year),
MIGRATION_STATE = OUTBOUND
) )

-> Executed below coommand to insert an row,

insert into Table1 values(replicate(‘A’,25),1980)

-> Now checking the Stretch monitor,

Blog6_3.PNG

-> It is clear from the screenshot above that the new filter function has taken effect.

Thank You,
Vivek Janakiraman

Disclaimer:
The views expressed on this blog are mine alone and do not reflect the views of my company or anyone else. All postings on this blog are provided “AS IS” with no warranties, and confers no rights.

Advertisements

Disabling Stretch Database

-> Please refer “https://jbswiki.com/2017/06/15/stretch-database-in-sql-server-2016/” if you want to setup Stretch database from scratch. In this article, I will be disabling stretch database on the SQL server instance.

-> Executing the below query to see what all tables are part of stretch database,

select object_name (object_id),* from sys.remote_data_archive_tables

Blog5_1

-> We must first disable the stretch for each individual tables before disabling it for the database.

-> Disabling it for Table1 by migrating the data back from Azure database to On-Premise database. Please note that migrating the data from azure will involve extra cost.

Blog5_2.PNG

-> Make sure we have sufficient space on the data drive to bring back the data and then click on “Yes”.

Blog5_3.PNG

Blog5_4.PNG

-> Thats quick. Usually it takes time for migrating, if the data involved is too huge.

-> Executing the below query to see what all tables are part of stretch database. We dont see Table1 anymore,

select object_name (object_id),* from sys.remote_data_archive_tables

Blog5_5.PNG

-> Checking the execution plan after executing the below query. The “Remote Query” is no longer present and the “actual number of rows” now is 50000.

select sname, year, count(*)
from Table1
group by sname,year

Blog5_6.PNG

-> Disabling stretch feature for Table2 by selecting option “Leave Data in Azure”.

Blog5_7.PNG

Blog5_8

Blog5_9

Blog5_10

-> Executing the below query to see what all tables are part of stretch database. We dont see any tables now,

select object_name (object_id),* from sys.remote_data_archive_tables

Blog5_11

-> Checking the execution plan after executing the below query. The “Remote Query” is no longer present and the “actual number of rows” now is 10000. It is clear that all the rows in azure is not migrated back to On-Premise database.

select sname, year, count(*)
from Table2
group by sname,year

Blog5_13

Blog5_12

-> Now that the tables are done, I will disable the stretch feature for the database,

Blog5_14.PNG

Blog5_15

Blog5_16

Thank You,
Vivek Janakiraman

Disclaimer:
The views expressed on this blog are mine alone and do not reflect the views of my company or anyone else. All postings on this blog are provided “AS IS” with no warranties, and confers no rights.

Adding a table to the stretch database

-> Please refer “https://jbswiki.com/2017/06/15/stretch-database-in-sql-server-2016/” if you want to setup Stretch database from scratch. In this article, I will be adding table Table3 to the existing stretch database.

-> Let’s create the required objects for this demo on the already configured Stretch database JB_StretchDB,

use JB_StretchDB
go
create table Table3(
sno int primary key identity (1,1),
sname varchar(255),
Year int)

-> Populating tables Table3 with some data,

set nocount on
insert into Table3 values(replicate(‘A’,25),1980)
go 10000
insert into Table3 values(replicate(‘B’,25),1981)
go 10000
insert into Table3 values(replicate(‘C’,25),1982)
go 10000
insert into Table3 values(replicate(‘D’,25),1983)
go 10000
insert into Table3 values(replicate(‘E’,25),1984)
go 10000

-> Looking at the data,

select sname, year, count(*)
from Table3
group by sname,year

Blog4_1

-> Select the table that will be added to stretch database and perform as shown in the screenshot,

Blog4_2

Blog4_3

-> Click on “Entire Table” under “Migrate”.

Blog4_4

Blog4_5.PNG

Blog4_6.PNG

Blog4_7

Blog4_8.PNG

-> Monitoring the status for table3,

Blog4_9.PNG

Blog4_10.PNG

-> Executing the below query and looking at the execution plan. The execution plan shows us that we are querying 10000 rows from the On-Premise database and 40000 rows from the Azure database.

select sname, year, count(*)
from Table3
group by sname,year

Blog4_11.PNG

Thank You,
Vivek Janakiraman

Disclaimer:
The views expressed on this blog are mine alone and do not reflect the views of my company or anyone else. All postings on this blog are provided “AS IS” with no warranties, and confers no rights.

Stretch Database in SQL Server 2016

We can use the stretch database to place our less-frequently accessed data securely in Microsoft Azure.

Benefits

-> Cost effective – The less-frequently accessed data can be stored on cost efficient and secure Microsoft Azure than storing them on costly On-premise storage.

-> No changes in queries – The queries that access a stretch database can be same and no additional logic is required to incorporate the data moved to Microsoft Azure.

-> Database Maintenance – Database maintenance will be comparably faster as we will touch only the frequently accessed data.

-> Faster queries – Queries touching frequently accessed database will be comparably faster with fewer IO and increased performance.

Blog1_Azure_visio

-> Let’s create the required database and objects for this demo,

create database JB_StretchDB
go
use JB_StretchDB
go
create table Table1(
sno int primary key identity (1,1),
sname varchar(255),
Year int)

-> Populating table Table1 with some data,

set nocount on
insert into Table1 values(replicate(‘A’,25),1980)
go 10000
insert into Table1 values(replicate(‘B’,25),1981)
go 10000
insert into Table1 values(replicate(‘C’,25),1982)
go 10000
insert into Table1 values(replicate(‘D’,25),1983)
go 10000
insert into Table1 values(replicate(‘E’,25),1984)
go 10000

-> Looking at the data,

select sname, year, count(*)
from Table1
group by sname,year

Blog3_1

-> The query completes in 60 MS.

Blog3_2

-> Select the database that will be part of stretch database,

Blog3_3.png

Blog3_4.png

-> Select the table and click on “Entire Table” under “Migrate”.

Blog3_5.png

-> You can move all the data to cloud or a certain number of rows. I will select “Choose Rows” in the below screen as I want the latest data in On-Premise database and just the old data’s be moved to cloud.

Blog3_6.png

Blog3_7.png

-> I already have a Microsoft azure instance created. Hence I will select “Existing server” and point the instance from the drop down box. You can create one yourself by using the article https://jbswiki.com/2017/05/31/creating-a-sql-database-in-microsoft-azure-portal/ or you can select “Create new server” and proceed.

Blog3_8.png

-> Provide the password for the database master key.

Blog3_9

Blog3_10

Blog3_11

Blog3_12

-> This is what I see in the SSMS after setting up the stretch database.

Blog3_13.png

-> Executing the below query and checking the execution plan. The query actually took 19 seconds.

Blog3_14.png

-> It is clear from the screenshot that we are spending most of our time on the remote query as we are querying close to 40000 rows from Azure database.

Blog3_15.png

-> The performance is not that great, but we create stretch database just to move data to cloud that are not frequently accessed.

Thank You,
Vivek Janakiraman

Disclaimer:
The views expressed on this blog are mine alone and do not reflect the views of my company or anyone else. All postings on this blog are provided “AS IS” with no warranties, and confers no rights.