Gegevensverzameling op de SQL Server-databasegroei automatiseren?

Inleiding

Vaak is het nodig om de groei van alle tabellen en bestanden van alle databases te beheersen.

In dit artikel gaan we een voorbeeld onderzoeken van het automatiseren van gegevensverzameling over de groei van SQL Server-databasetabellen en -bestanden.

Oplossing

1. Maak een weergave over de grootte van alle tabellen voor elke database

   USE [DATABASE_NAME]
   GO
   
   SET ANSI_NULLS ON
   GO
   
   SET QUOTED_IDENTIFIER ON
   GO
   
   
   CREATE view [inf].[vTableSize] as
   with pagesizeKB as (
   	SELECT low / 1024 as PageSizeKB
   	FROM master.dbo.spt_values
   	WHERE number = 1 AND type = 'E'
   )
   ,f_size as (
   	select p.[object_id], 
   		   sum([total_pages]) as TotalPageSize,
   		   sum([used_pages])  as UsedPageSize,
   		   sum([data_pages])  as DataPageSize
   	from sys.partitions p join sys.allocation_units a on p.partition_id = a.container_id
   	left join sys.internal_tables it on p.object_id = it.object_id
   	WHERE OBJECTPROPERTY(p.[object_id], N'IsUserTable') = 1
   	group by p.[object_id]
   )
   ,tbl as (
   	SELECT
   	  t.[schema_id],
   	  t.[object_id],
   	  i1.rowcnt as CountRows,
   	  (COALESCE(SUM(i1.reserved), 0) + COALESCE(SUM(i2.reserved), 0)) * (select top(1) PageSizeKB from pagesizeKB) as ReservedKB,
   	  (COALESCE(SUM(i1.dpages), 0) + COALESCE(SUM(i2.used), 0)) * (select top(1) PageSizeKB from pagesizeKB) as DataKB,
   	  ((COALESCE(SUM(i1.used), 0) + COALESCE(SUM(i2.used), 0))
   	    - (COALESCE(SUM(i1.dpages), 0) + COALESCE(SUM(i2.used), 0))) * (select top(1) PageSizeKB from pagesizeKB) as IndexSizeKB,
   	  ((COALESCE(SUM(i1.reserved), 0) + COALESCE(SUM(i2.reserved), 0))
   	    - (COALESCE(SUM(i1.used), 0) + COALESCE(SUM(i2.used), 0))) * (select top(1) PageSizeKB from pagesizeKB) as UnusedKB
   	FROM sys.tables as t
   	LEFT OUTER JOIN sysindexes as i1 ON i1.id = t.[object_id] AND i1.indid < 2
   	LEFT OUTER JOIN sysindexes as i2 ON i2.id = t.[object_id] AND i2.indid = 255
   	WHERE OBJECTPROPERTY(t.[object_id], N'IsUserTable') = 1
   	OR (OBJECTPROPERTY(t.[object_id], N'IsView') = 1 AND OBJECTPROPERTY(t.[object_id], N'IsIndexed') = 1)
   	GROUP BY t.[schema_id], t.[object_id], i1.rowcnt
   )
   SELECT
     @@Servername AS Server,
     DB_NAME() AS DBName,
     SCHEMA_NAME(t.[schema_id]) as SchemaName,
     OBJECT_NAME(t.[object_id]) as TableName,
     t.CountRows,
     t.ReservedKB,
     t.DataKB,
     t.IndexSizeKB,
     t.UnusedKB,
     f.TotalPageSize*(select top(1) PageSizeKB from pagesizeKB) as TotalPageSizeKB,
     f.UsedPageSize*(select top(1) PageSizeKB from pagesizeKB) as UsedPageSizeKB,
     f.DataPageSize*(select top(1) PageSizeKB from pagesizeKB) as DataPageSizeKB
   FROM f_size as f
   inner join tbl as t on t.[object_id]=f.[object_id]
   
   GO

2. Maak een specifieke database en bepaal een tabel voor het opslaan van informatie over de groei van alle databasetabellen:

   USE [DATABASE_NAME]
   GO
   
   SET ANSI_NULLS ON
   GO
   
   SET QUOTED_IDENTIFIER ON
   GO
   
   SET ANSI_PADDING ON
   GO
   
   CREATE TABLE [srv].[TableStatistics](
   	[Row_GUID] [uniqueidentifier] NOT NULL CONSTRAINT [DF_TableStatistics_Row_GUID]  DEFAULT (newid()),
   	[ServerName] [nvarchar](255) NOT NULL,
   	[DBName] [nvarchar](255) NOT NULL,
   	[SchemaName] [nvarchar](255) NOT NULL,
   	[TableName] [nvarchar](255) NOT NULL,
   	[CountRows] [bigint] NOT NULL,
   	[DataKB] [int] NOT NULL,
   	[IndexSizeKB] [int] NOT NULL,
   	[UnusedKB] [int] NOT NULL,
   	[ReservedKB] [int] NOT NULL,
   	[InsertUTCDate] [datetime] NOT NULL CONSTRAINT [DF_TableStatistics_InsertUTCDate]  DEFAULT (getutcdate()),
   	[Date]  AS (CONVERT([date],[InsertUTCDate])) PERSISTED,
   	[CountRowsBack] [bigint] NULL,
   	[CountRowsNext] [bigint] NULL,
   	[DataKBBack] [int] NULL,
   	[DataKBNext] [int] NULL,
   	[IndexSizeKBBack] [int] NULL,
   	[IndexSizeKBNext] [int] NULL,
   	[UnusedKBBack] [int] NULL,
   	[UnusedKBNext] [int] NULL,
   	[ReservedKBBack] [int] NULL,
   	[ReservedKBNext] [int] NULL,
   	[AvgCountRows]  AS ((([CountRowsBack]+[CountRows])+[CountRowsNext])/(3)) PERSISTED,
   	[AvgDataKB]  AS ((([DataKBBack]+[DataKB])+[DataKBNext])/(3)) PERSISTED,
   	[AvgIndexSizeKB]  AS ((([IndexSizeKBBack]+[IndexSizeKB])+[IndexSizeKBNext])/(3)) PERSISTED,
   	[AvgUnusedKB]  AS ((([UnusedKBBack]+[UnusedKB])+[UnusedKBNext])/(3)) PERSISTED,
   	[AvgReservedKB]  AS ((([ReservedKBBack]+[ReservedKB])+[ReservedKBNext])/(3)) PERSISTED,
   	[DiffCountRows]  AS (([CountRowsNext]+[CountRowsBack])-(2)*[CountRows]) PERSISTED,
   	[DiffDataKB]  AS (([DataKBNext]+[DataKBBack])-(2)*[DataKB]) PERSISTED,
   	[DiffIndexSizeKB]  AS (([IndexSizeKBNext]+[IndexSizeKBBack])-(2)*[IndexSizeKB]) PERSISTED,
   	[DiffUnusedKB]  AS (([UnusedKBNext]+[UnusedKBBack])-(2)*[UnusedKB]) PERSISTED,
   	[DiffReservedKB]  AS (([ReservedKBNext]+[ReservedKBBack])-(2)*[ReservedKB]) PERSISTED,
   	[TotalPageSizeKB] [int] NULL,
   	[TotalPageSizeKBBack] [int] NULL,
   	[TotalPageSizeKBNext] [int] NULL,
   	[UsedPageSizeKB] [int] NULL,
   	[UsedPageSizeKBBack] [int] NULL,
   	[UsedPageSizeKBNext] [int] NULL,
   	[DataPageSizeKB] [int] NULL,
   	[DataPageSizeKBBack] [int] NULL,
   	[DataPageSizeKBNext] [int] NULL,
   	[AvgDataPageSizeKB]  AS ((([DataPageSizeKBBack]+[DataPageSizeKB])+[DataPageSizeKBNext])/(3)) PERSISTED,
   	[AvgUsedPageSizeKB]  AS ((([UsedPageSizeKBBack]+[UsedPageSizeKB])+[UsedPageSizeKBNext])/(3)) PERSISTED,
   	[AvgTotalPageSizeKB]  AS ((([TotalPageSizeKBBack]+[TotalPageSizeKB])+[TotalPageSizeKBNext])/(3)) PERSISTED,
   	[DiffDataPageSizeKB]  AS (([DataPageSizeKBNext]+[DataPageSizeKBBack])-(2)*[DataPageSizeKB]) PERSISTED,--shows as the casting is changed
   	[DiffUsedPageSizeKB]  AS (([UsedPageSizeKBNext]+[UsedPageSizeKBBack])-(2)*[UsedPageSizeKB]) PERSISTED,--shows as the casting is changed
   	[DiffTotalPageSizeKB]  AS (([TotalPageSizeKBNext]+[TotalPageSizeKBBack])-(2)*[TotalPageSizeKB]) PERSISTED,--shows as the casting is changed
    CONSTRAINT [PK_TableStatistics] PRIMARY KEY CLUSTERED 
   (
   	[Row_GUID] ASC
   )WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON) ON [PRIMARY]
   ) ON [PRIMARY]
   
   GO
   
   SET ANSI_PADDING ON
   GO

   TotalPageSizeKB geeft de tabelgrootte aan.

   Het totaal van TotalPageSizeKB van alle databasetabellen + de grootte van systeemtabellen =de grootte van databasegegevens.

3. Bepaal de procedure voor het verzamelen van de informatie:

   USE [DATABASE_NAME]
   GO
   
   SET ANSI_NULLS ON
   GO
   
   SET QUOTED_IDENTIFIER ON
   GO
   
   CREATE PROCEDURE [srv].[InsertTableStatistics]
   AS
   BEGIN
   	SET NOCOUNT ON;
   	SET TRANSACTION ISOLATION LEVEL READ UNCOMMITTED;
   
   	declare @dt date=CAST(GetUTCDate() as date);
       declare @dbs nvarchar(255);
   	declare @sql nvarchar(max);
   
   	select [name]
   	into #dbs
   	from sys.databases;
   
   	while(exists(select top(1) 1 from #dbs))
   	begin
   		select top(1)
   		@dbs=[name]
   		from #dbs;
   
   		set @sql=
   		N'INSERT INTO [srv].[TableStatistics]
   	         ([ServerName]
   			   ,[DBName]
   	         ,[SchemaName]
   	         ,[TableName]
   	         ,[CountRows]
   	         ,[DataKB]
   	         ,[IndexSizeKB]
   	         ,[UnusedKB]
   	         ,[ReservedKB]
   			 ,[TotalPageSizeKB]
   			 ,[UsedPageSizeKB]
   			 ,[DataPageSizeKB])
   	   SELECT [Server]
   		  ,[DBName]
   	         ,[SchemaName]
   	         ,[TableName]
   	         ,[CountRows]
   	         ,[DataKB]
   	         ,[IndexSizeKB]
   	         ,[UnusedKB]
   	         ,[ReservedKB]
   			 ,[TotalPageSizeKB]
   			 ,[UsedPageSizeKB]
   			 ,[DataPageSizeKB]
   		FROM ['[email protected]+'].[inf].[vTableSize];';
   
   		exec sp_executesql @sql;
   
   		delete from #dbs
   		where [name][email protected];
   	end
   
   	drop table #dbs;
   
   	declare @dt_back date=CAST(DateAdd(day,-1,@dt) as date);
   
   	;with tbl1 as (
   		select [Date], 
   			   [CountRows],
   			   [DataKB],
   			   [IndexSizeKB],
   			   [UnusedKB],
   			   [ReservedKB],
   			   [ServerName], 
   			   [DBName], 
   			   [SchemaName], 
   			   [TableName],
   			   [TotalPageSizeKB],
   			   [UsedPageSizeKB],
   			   [DataPageSizeKB]
   		from [srv].[TableStatistics]
   		where [Date][email protected]_back
   	)
   	, tbl2 as (
   		select [Date], 
   			   [CountRows], 
   			   [CountRowsBack],
   			   [DataKBBack],
   			   [IndexSizeKBBack],
   			   [UnusedKBBack],
   			   [ReservedKBBack],
   			   [ServerName], 
   			   [DBName], 
   			   [SchemaName], 
   			   [TableName],
   			   [TotalPageSizeKBBack],
   			   [UsedPageSizeKBBack],
   			   [DataPageSizeKBBack]
   		from [srv].[TableStatistics]
   		where [Date][email protected]
   	)
   	update t2
   	set t2.[CountRowsBack]		=t1.[CountRows],
   		t2.[DataKBBack]			=t1.[DataKB],
   		t2.[IndexSizeKBBack]	=t1.[IndexSizeKB],
   		t2.[UnusedKBBack]		=t1.[UnusedKB],
   		t2.[ReservedKBBack]		=t1.[ReservedKB],
   		t2.[TotalPageSizeKBBack]=t1.[TotalPageSizeKB],
   		t2.[UsedPageSizeKBBack]	=t1.[UsedPageSizeKB],
   		t2.[DataPageSizeKBBack]	=t1.[DataPageSizeKB]
   	from tbl1 as t1
   	inner join tbl2 as t2 on t1.[Date]=DateAdd(day,-1,t2.[Date])
   	and t1.[ServerName]=t2.[ServerName]
   	and t1.[DBName]=t2.[DBName]
   	and t1.[SchemaName]=t2.[SchemaName]
   	and t1.[TableName]=t2.[TableName];
   
   	;with tbl1 as (
   		select [Date], 
   			   [CountRows], 
   			   [CountRowsNext],
   			   [DataKBNext],
   			   [IndexSizeKBNext],
   			   [UnusedKBNext],
   			   [ReservedKBNext],
   			   [ServerName], 
   			   [DBName], 
   			   [SchemaName], 
   			   [TableName],
   			   [TotalPageSizeKBNext],
   			   [UsedPageSizeKBNext],
   			   [DataPageSizeKBNext]
   		from [srv].[TableStatistics]
   		where [Date][email protected]_back
   	)
   	, tbl2 as (
   		select [Date], 
   			   [CountRows],
   			   [DataKB],
   			   [IndexSizeKB],
   			   [UnusedKB],
   			   [ReservedKB],
   			   [ServerName], 
   			   [DBName], 
   			   [SchemaName], 
   			   [TableName],
   			   [TotalPageSizeKB],
   			   [UsedPageSizeKB],
   			   [DataPageSizeKB]
   		from [srv].[TableStatistics]
   		where [Date][email protected]
   	)
   	update t1
   	set t1.[CountRowsNext]		=t2.[CountRows],
   		t1.[DataKBNext]			=t2.[DataKB],
   		t1.[IndexSizeKBNext]	=t2.[IndexSizeKB],
   		t1.[UnusedKBNext]		=t2.[UnusedKB],
   		t1.[ReservedKBNext]		=t2.[ReservedKB],
   		t1.[TotalPageSizeKBNext]=t2.[TotalPageSizeKB],
   		t1.[UsedPageSizeKBNext]	=t2.[UsedPageSizeKB],
   		t1.[DataPageSizeKBNext]	=t2.[DataPageSizeKB]
   	from tbl1 as t1
   	inner join tbl2 as t2 on t1.[Date]=DateAdd(day,-1,t2.[Date])
   	and t1.[ServerName]=t2.[ServerName]
   	and t1.[DBName]=t2.[DBName]
   	and t1.[SchemaName]=t2.[SchemaName]
   	and t1.[TableName]=t2.[TableName];
   END
   GO

   Deze oplossing kan worden aangepast om gegevens te verzamelen over de grootte van tabellen van alle databases van alle vereiste instanties van MS SQL Server.

4. Definieer de weergave op basis van de verzamelde informatie:

   USE [DATABASE_NAME]
   GO
   
   SET ANSI_NULLS ON
   GO
   
   SET QUOTED_IDENTIFIER ON
   GO
   
   create view [srv].[vTableStatisticsShort] as 
   with d as (select DateAdd(day,-1,max([Date])) as [Date] from [srv].[TableStatistics])
   SELECT t.[ServerName]
         ,t.[DBName]
         ,t.[SchemaName]
         ,t.[TableName]
         ,t.[CountRows]
         ,t.[DataKB]
         ,t.[IndexSizeKB]
         ,t.[UnusedKB]
         ,t.[ReservedKB]
         ,t.[InsertUTCDate]
         ,t.[Date]
         ,t.[CountRowsBack]
         ,t.[CountRowsNext]
         ,t.[DataKBBack]
         ,t.[DataKBNext]
         ,t.[IndexSizeKBBack]
         ,t.[IndexSizeKBNext]
         ,t.[UnusedKBBack]
         ,t.[UnusedKBNext]
         ,t.[ReservedKBBack]
         ,t.[ReservedKBNext]
         ,t.[AvgCountRows]
         ,t.[AvgDataKB]
         ,t.[AvgIndexSizeKB]
         ,t.[AvgUnusedKB]
         ,t.[AvgReservedKB]
         ,t.[DiffCountRows]
         ,t.[DiffDataKB]
         ,t.[DiffIndexSizeKB]
         ,t.[DiffUnusedKB]
         ,t.[DiffReservedKB]
         ,t.[TotalPageSizeKB]
         ,t.[TotalPageSizeKBBack]
         ,t.[TotalPageSizeKBNext]
         ,t.[UsedPageSizeKB]
         ,t.[UsedPageSizeKBBack]
         ,t.[UsedPageSizeKBNext]
         ,t.[DataPageSizeKB]
         ,t.[DataPageSizeKBBack]
         ,t.[DataPageSizeKBNext]
         ,t.[AvgDataPageSizeKB]
         ,t.[AvgUsedPageSizeKB]
         ,t.[AvgTotalPageSizeKB]
         ,t.[DiffDataPageSizeKB]
         ,t.[DiffUsedPageSizeKB]
         ,t.[DiffTotalPageSizeKB]
     FROM d
     inner join [SRV].[srv].[TableStatistics] as t on d.[Date]=t.[Date]
     where t.[CountRowsBack] is not null
   	and	t.[CountRowsNext] is not null
   GO

   Hierbij wil ik uw aandacht vestigen op Diff. Als het hoger is dan 0, betekent dit dat de tafel elke dag sneller groeit.

   De collectie zou eens per 24 uur moeten plaatsvinden.

   Op dezelfde manier kunnen we het verzamelen van de bestandsgroei van alle databases automatiseren met behulp van de volgende weergave:

   USE [DATABASE_NAME]
   GO
   
   SET ANSI_NULLS ON
   GO
   
   SET QUOTED_IDENTIFIER ON
   GO
   
   select	t2.[DB_Name] as [DBName]
   			,t1.FileId 
   			,t1.NumberReads
   			,t1.BytesRead
   			,t1.IoStallReadMS
   			,t1.NumberWrites
   			,t1.BytesWritten
   			,t1.IoStallWriteMS 
   			,t1.IoStallMS
   			,t1.BytesOnDisk
   			,t1.[TimeStamp]
   			,t1.FileHandle
   			,t2.[Type_desc]
   			,t2.[FileName]
   			,t2.[Drive]
   			,t2.[Physical_Name]
   			,t2.[Ext]
   			,t2.[CountPage]
   			,t2.[SizeMb]
   			,t2.[SizeGb]
   			,t2.[Growth]
   			,t2.[GrowthMb]
   			,t2.[GrowthGb]
   			,t2.[GrowthPercent]
   			,t2.[is_percent_growth]
   			,t2.[database_id]
   			,t2.[State]
   			,t2.[StateDesc]
   			,t2.[IsMediaReadOnly]
   			,t2.[IsReadOnly]
   			,t2.[IsSpace]
   			,t2.[IsNameReserved]
   			,t2.[CreateLsn]
   			,t2.[DropLsn]
   			,t2.[ReadOnlyLsn]
   			,t2.[ReadWriteLsn]
   			,t2.[DifferentialBaseLsn]
   			,t2.[DifferentialBaseGuid]
   			,t2.[DifferentialBaseTime]
   			,t2.[RedoStartLsn]
   			,t2.[RedoStartForkGuid]
   			,t2.[RedoTargetLsn]
   			,t2.[RedoTargetForkGuid]
   			,t2.[BackupLsn]
   from fn_virtualfilestats(NULL, NULL) as t1
   inner join [inf].[ServerDBFileInfo] as t2 on t1.[DbId]=t2.[database_id] and t1.[FileId]=t2.[File_Id]
   GO

Resultaat

In dit artikel hebben we het voorbeeld onderzocht van het automatiseren van gegevensverzameling over de grootte en groei van alle SQL Server-tabellen en -bestanden van alle databases. Het geeft ons de volledige controle over het wijzigen van de grootte van zowel de databasebestanden als de tabellen, evenals het nemen van tijdige maatregelen om de tabel of het bestand te verkleinen, het gegevensopslagapparaat te vergroten of informatie op te splitsen in meerdere gegevensopslagapparaten.