【发布时间】：2011-08-22 04:02:24
【问题描述】：
我有 MS SQL 2008 R2，500 个数据库。 查询所有数据库大小的最有效、最简单和“现代”的方法是什么。
输出应该有列：
- 数据库名称
- 数据文件大小
- 日志文件大小
【问题讨论】：
标签： sql sql-server tsql sql-server-2008-r2
【发布时间】：2011-08-22 04:02:24
【问题描述】：
with fs
as
(
select database_id, type, size * 8.0 / 1024 size
from sys.master_files
)
select
name,
(select sum(size) from fs where type = 0 and fs.database_id = db.database_id) DataFileSizeMB,
(select sum(size) from fs where type = 1 and fs.database_id = db.database_id) LogFileSizeMB
from sys.databases db
【讨论】：
-
@Derek Smith - 大小列以页为单位。一页的大小是 8kB，因此公式 SizeInMb = PagesCount * 8 / 1024。我一直认为 1MB = 1024kB。我认为即使是微软也没有足够的力量来改变这一点:)。
-
这个答案由于某种原因在我们的系统上报告了错误的 tempdb 值。请参阅this related question。
我不知道你所说的效率到底是什么意思，但这很简单，对我有用：
SELECT
DB_NAME(db.database_id) DatabaseName,
(CAST(mfrows.RowSize AS FLOAT)*8)/1024 RowSizeMB,
(CAST(mflog.LogSize AS FLOAT)*8)/1024 LogSizeMB,
(CAST(mfstream.StreamSize AS FLOAT)*8)/1024 StreamSizeMB,
(CAST(mftext.TextIndexSize AS FLOAT)*8)/1024 TextIndexSizeMB
FROM sys.databases db
LEFT JOIN (SELECT database_id, SUM(size) RowSize FROM sys.master_files WHERE type = 0 GROUP BY database_id, type) mfrows ON mfrows.database_id = db.database_id
LEFT JOIN (SELECT database_id, SUM(size) LogSize FROM sys.master_files WHERE type = 1 GROUP BY database_id, type) mflog ON mflog.database_id = db.database_id
LEFT JOIN (SELECT database_id, SUM(size) StreamSize FROM sys.master_files WHERE type = 2 GROUP BY database_id, type) mfstream ON mfstream.database_id = db.database_id
LEFT JOIN (SELECT database_id, SUM(size) TextIndexSize FROM sys.master_files WHERE type = 4 GROUP BY database_id, type) mftext ON mftext.database_id = db.database_id
结果如下：
DatabaseName RowSizeMB LogSizeMB StreamSizeMB TextIndexSizeMB
------------- --------- --------- ------------ ---------------
master 4 1.25 NULL NULL
model 2.25 0.75 NULL NULL
msdb 14.75 8.1875 NULL NULL
tempdb 8 0.5 NULL NULL
注意：灵感来自this article
【讨论】：
-
我使用这个在每一列上都得到 NULL 值？知道为什么吗？
-
如果 sys.master_files 为空，则可能是权限。最小权限是 CREATE DATABASE、ALTER ANY DATABASE 或 VIEW ANY DEFINITION（参考：BOL sys.master_files）
这是一个简单、快速且可靠的查询，它将以漂亮、易于阅读的输出提供所有数据库和日志文件的名称、大小以及数据库状态（例如 ONLINE）：
SELECT
D.name,
F.Name AS FileType,
F.physical_name AS PhysicalFile,
F.state_desc AS OnlineStatus,
CAST(F.size AS bigint) * 8*1024 AS SizeInBytes,
CAST((F.size*8.0)/1024/1024 AS decimal(18,3)) AS SizeInGB
FROM
sys.master_files F
INNER JOIN sys.databases D ON D.database_id = F.database_id
ORDER BY SizeInBytes desc
【讨论】：
按数据库总大小排序 Desc
SELECT
DB_NAME(db.database_id) DatabaseName,
(CAST(mfrows.RowSize AS FLOAT)*8)/1024 RowSizeMB,
(CAST(mflog.LogSize AS FLOAT)*8)/1024 LogSizeMB,
(CAST(mfrows.RowSize AS FLOAT)*8)/1024/1024+(CAST(mflog.LogSize AS FLOAT)*8)/1024/1024 DBSizeG,
(CAST(mfstream.StreamSize AS FLOAT)*8)/1024 StreamSizeMB,
(CAST(mftext.TextIndexSize AS FLOAT)*8)/1024 TextIndexSizeMB
FROM sys.databases db
LEFT JOIN (SELECT database_id,
SUM(size) RowSize
FROM sys.master_files
WHERE type = 0
GROUP BY database_id, type) mfrows
ON mfrows.database_id = db.database_id
LEFT JOIN (SELECT database_id,
SUM(size) LogSize
FROM sys.master_files
WHERE type = 1
GROUP BY database_id, type) mflog
ON mflog.database_id = db.database_id
LEFT JOIN (SELECT database_id,
SUM(size) StreamSize
FROM sys.master_files
WHERE type = 2
GROUP BY database_id, type) mfstream
ON mfstream.database_id = db.database_id
LEFT JOIN (SELECT database_id,
SUM(size) TextIndexSize
FROM sys.master_files
WHERE type = 4
GROUP BY database_id, type) mftext
ON mftext.database_id = db.database_id
ORDER BY 4 DESC
【讨论】：
一切似乎都过于复杂！还是我错过了什么？
当然，您只需要这样的东西：
select d.name, case when m.type = 0 then 'Data' else 'Log' end, m.size * 8 / 1024
from sys.master_files m JOIN sys.databases d ON d.database_id = m.database_id
或者如果你不想要日志：
select d.name, m.size * 8 / 1024
from sys.master_files m JOIN sys.databases d ON d.database_id = m.database_id and m.type =0
【讨论】：
SELECT
DB.name,
SUM(CASE WHEN type = 0 THEN MF.size * 8 / 1024 ELSE 0 END) AS DataFileSizeMB,
SUM(CASE WHEN type = 1 THEN MF.size * 8 / 1024 ELSE 0 END) AS LogFileSizeMB
FROM
sys.master_files MF
JOIN sys.databases DB ON DB.database_id = MF.database_id
GROUP BY DB.name
ORDER BY DataFileSizeMB DESC
【讨论】：
-
GROUP BY DB.name, DB.database_id ORDER BY DB.database_id保留表格顺序。
请查找更多详细信息或从以下链接下载脚本 https://gallery.technet.microsoft.com/SIZE-OF-ALL-DATABASES-IN-0337f6d5#content
DECLARE @spacetable table
(
database_name varchar(50) ,
total_size_data int,
space_util_data int,
space_data_left int,
percent_fill_data float,
total_size_data_log int,
space_util_log int,
space_log_left int,
percent_fill_log char(50),
[total db size] int,
[total size used] int,
[total size left] int
)
insert into @spacetable
EXECUTE master.sys.sp_MSforeachdb 'USE [?];
select x.[DATABASE NAME],x.[total size data],x.[space util],x.[total size data]-x.[space util] [space left data],
x.[percent fill],y.[total size log],y.[space util],
y.[total size log]-y.[space util] [space left log],y.[percent fill],
y.[total size log]+x.[total size data] ''total db size''
,x.[space util]+y.[space util] ''total size used'',
(y.[total size log]+x.[total size data])-(y.[space util]+x.[space util]) ''total size left''
from (select DB_NAME() ''DATABASE NAME'',
sum(size*8/1024) ''total size data'',sum(FILEPROPERTY(name,''SpaceUsed'')*8/1024) ''space util''
,case when sum(size*8/1024)=0 then ''less than 1% used'' else
substring(cast((sum(FILEPROPERTY(name,''SpaceUsed''))*1.0*100/sum(size)) as CHAR(50)),1,6) end ''percent fill''
from sys.master_files where database_id=DB_ID(DB_NAME()) and type=0
group by type_desc ) as x ,
(select
sum(size*8/1024) ''total size log'',sum(FILEPROPERTY(name,''SpaceUsed'')*8/1024) ''space util''
,case when sum(size*8/1024)=0 then ''less than 1% used'' else
substring(cast((sum(FILEPROPERTY(name,''SpaceUsed''))*1.0*100/sum(size)) as CHAR(50)),1,6) end ''percent fill''
from sys.master_files where database_id=DB_ID(DB_NAME()) and type=1
group by type_desc )y'
select * from @spacetable
order by database_name
【讨论】：
我最近看到这个页面正在寻找类似的东西。 万一有人遇到这种情况并且只对核心用户数据库感兴趣，您可以使用类似这样的东西，不包括 Master、msdb...
SELECT
DB_NAME(db.database_id) DatabaseName,
(CAST(mfrows.RowSize AS FLOAT)*8)/1024 RowSizeMB,
(CAST(mflog.LogSize AS FLOAT)*8)/1024 LogSizeMB
FROM sys.databases db
LEFT JOIN (SELECT database_id, SUM(size) RowSize FROM sys.master_files WHERE type = 0 GROUP BY database_id, type) mfrows ON mfrows.database_id = db.database_id
LEFT JOIN (SELECT database_id, SUM(size) LogSize FROM sys.master_files WHERE type = 1 GROUP BY database_id, type) mflog ON mflog.database_id = db.database_id
where DB_NAME(db.database_id) not like 'master'
and DB_NAME(db.database_id) not like 'msdb'
and DB_NAME(db.database_id) not like 'model'
and DB_NAME(db.database_id) not like 'tempdb'
and DB_NAME(db.database_id) not like 'Northwind'
and DB_NAME(db.database_id) not like 'ReportServer'
order by DB_NAME(db.database_id)
【讨论】：
IF OBJECT_ID('tempdb.dbo.#space') IS NOT NULL
DROP TABLE #space
CREATE TABLE #space (
database_id INT PRIMARY KEY
, data_used_size DECIMAL(18,2)
, log_used_size DECIMAL(18,2)
)
DECLARE @SQL NVARCHAR(MAX)
SELECT @SQL = STUFF((
SELECT '
USE [' + d.name + ']
INSERT INTO #space (database_id, data_used_size, log_used_size)
SELECT
DB_ID()
, SUM(CASE WHEN [type] = 0 THEN space_used END)
, SUM(CASE WHEN [type] = 1 THEN space_used END)
FROM (
SELECT s.[type], space_used = SUM(FILEPROPERTY(s.name, ''SpaceUsed'') * 8. / 1024)
FROM sys.database_files s
GROUP BY s.[type]
) t;'
FROM sys.databases d
WHERE d.[state] = 0
FOR XML PATH(''), TYPE).value('.', 'NVARCHAR(MAX)'), 1, 2, '')
EXEC sys.sp_executesql @SQL
SELECT
d.database_id
, d.name
, d.state_desc
, d.recovery_model_desc
, t.total_size
, t.data_size
, s.data_used_size
, t.log_size
, s.log_used_size
FROM (
SELECT
database_id
, log_size = CAST(SUM(CASE WHEN [type] = 1 THEN size END) * 8. / 1024 AS DECIMAL(18,2))
, data_size = CAST(SUM(CASE WHEN [type] = 0 THEN size END) * 8. / 1024 AS DECIMAL(18,2))
, total_size = CAST(SUM(size) * 8. / 1024 AS DECIMAL(18,2))
FROM sys.master_files
GROUP BY database_id
) t
JOIN sys.databases d ON d.database_id = t.database_id
LEFT JOIN #space s ON d.database_id = s.database_id
ORDER BY t.total_size DESC
【讨论】：
不要窃取你的答案并将其调整为分数或任何东西，但这是另一个分解：
select d.name,
sum(m0.size*8.0/1024) data_file_size_mb,
sum(m1.size*8.0/1024) log_file_size_mb
from sys.databases d
inner join sys.master_files m0 on m0.database_id = d.database_id
inner join sys.master_files m1 on m1.database_id = d.database_id
where m0.type = 0 and m1.type = 1
group by d.name, d.database_id
order by d.database_id
【讨论】：
-
Mihai Bejenariu 的 答案更接近我在这里的尝试。
简化和改进版：
SELECT
D.name,
CAST(SUM(F.size) AS bigint) * 8*1024 AS SizeInBytes,
CAST(SUM(F.size*8.0)/1024/1024 AS decimal(18,3)) AS SizeInGB
FROM
sys.master_files F
INNER JOIN sys.databases D ON D.database_id = F.database_id
GROUP BY D.name
ORDER BY SizeInBytes desc
【讨论】：
更好更简单的一个
SELECT [Database Name] = DB_NAME(database_id),
[Type] = CASE WHEN Type_Desc = 'ROWS' THEN 'Data File(s)'
WHEN Type_Desc = 'LOG' THEN 'Log File(s)'
ELSE Type_Desc END,
[Size in MB] = CAST( ((SUM(Size)* 8) / 1024.0) AS DECIMAL(18,2) )
FROM sys.master_files
--Uncomment if you need to query for a particular database
--WHERE database_id = DB_ID(‘Database Name’)
GROUP BY GROUPING SETS
(
(DB_NAME(database_id), Type_Desc),
(DB_NAME(database_id))
) ORDER BY DB_NAME(database_id), Type_Desc DESC
它将分别为您提供数据文件和日志文件的大小，如下所示
DatabaseName Type Size in MB
-------------------------------------------
FMS Data File(s) 23.00
FMS Log File(s) 1.50
PointOfSale Data File(s) 4.00
PointOfSale Log File(s) 1.25
Union2 Data File(s) 336.00
Union2 Log File(s) 1191.13
SurveyProject Data File(s) 4.00
SurveyProject Log File(s) 1.00
【讨论】：
-
这不是两个分组集的输出。
以下代码非常适合我。
SELECT
D.name As DbName,
F.Name AS FullDbName,
CASE WHEN F.type_desc='ROWS' THEN 'mdf' ELSE 'ldf' END AS FileType,
F.physical_name AS PhysicalFile,
CONVERT(DATE,D.create_date) AS CreationDate,
F.state_desc AS OnlineStatus,
CAST((F.size*8)/1024 AS VARCHAR(26)) + ' MB' AS FileSize_MB,
CAST(F.size*8 AS VARCHAR(32)) + ' Bytes' AS FileSize_Bytes,
CAST(CAST(ROUND((F.size*8)/(1024.0*1024.0),0) AS INT) AS VARCHAR(32)) + ' GB' AS FileSize_GB
FROM
sys.master_files F
INNER JOIN sys.databases D ON D.database_id = F.database_id
ORDER BY
D.name
【讨论】：
所有这些示例在我的大多数数据库服务器上都可以正常工作。 但是，如果您有 1 台具有多个实例的服务器，并且所有这些服务器都不会给您任何结果。
例如，第一个查询给出的结果如下：
name DataFileSizeMB LogFileSizeMB
master NULL NULL
tempdb NULL NULL
model NULL NULL
msdb NULL NULL
所以数据库是从sys.databases 中选择的，但是表sys.master_files 似乎是空的或没有给出结果。
即使是像select * from sys.master_files 这样的简单查询也有 0 条记录的结果。
您没有收到任何错误，但没有找到记录。
在我只有 1 个数据库实例的服务器上，这可以正常工作。
【讨论】：
-
欢迎来到 StackOverflow。我建议你下次改进你的答案，看看这里→How do I write a good answer?
-
如果您有新问题，请点击 按钮提出问题。如果有助于提供上下文，请包含指向此问题的链接。 - From Review
有时 SECURITY 问题会阻止询问所有 db，您需要使用 db 前缀逐个查询，对于这些情况，我创建了此动态查询
go
declare @Results table ([Name] nvarchar(max), [DataFileSizeMB] int, [LogFileSizeMB] int);
declare @QaQuery nvarchar(max)
declare @name nvarchar(max)
declare MY_CURSOR cursor
local static read_only forward_only
for
select name from master.dbo.sysdatabases where name not in ('master', 'tempdb', 'model', 'msdb', 'rdsadmin');
open MY_CURSOR
fetch next from MY_CURSOR into @name
while @@FETCH_STATUS = 0
begin
if(len(@name)>0)
begin
print @name + ' Column Exist'
set @QaQuery = N'select
'''+@name+''' as Name
,sum(case when type = 0 then size else 0 end) as DataFileSizeMB
,sum(case when type = 1 then size else 0 end) as LogFileSizeMB
from ['+@name+'].sys.database_files
group by replace(name, ''_log'', '''')';
insert @Results exec sp_executesql @QaQuery;
end
fetch next from MY_CURSOR into @name
end
close MY_CURSOR
deallocate MY_CURSOR
select * from @Results order by DataFileSizeMB desc
go
【讨论】：