人人都是 DBA（XI）I/O 信息收集腳本匯編

本文轉載自查看原文 2014-12-10 18:52 2306 T-SQL/ DBA/ SQL Server/ Troubleshooting/ Performance Tuning

什么？有個 SQL 執行了 8 秒！

哪里出了問題？臣妾不知道啊，得找 DBA 啊。

DBA 人呢？離職了！！擦！！！

程序員在無處尋求幫助時，就得想辦法自救，努力讓自己變成 "偽 DBA"。

索引

數據文件和日志文件位置和大小
查看指定數據庫文件的大小和可用空間
服務器 Disk 容量和掛載信息
查看 Disk 剩余空間
查詢數據庫設置的 Recovery Model
查看最近的 Full Backup 信息
獲取所有數據庫的 VLF 數量
SQL Server 的錯誤日志位置
查詢近期的 Error Log 信息
在錯誤日志中查詢 I/O 超過 15s 的請求
查詢 Disk 的性能指標
查看哪個數據庫文件 I/O 瓶頸最嚴重
按照 Write I/O 進行排名
獲取數據庫的 I/O 使用率
查看指定數據庫文件的 I/O 狀況
找出 I/O 平均使用最多的語句
查詢正在等待 I/O 的請求等待時間

數據文件和日志文件位置和大小

SELECT DB_NAME([database_id]) AS [Database Name]
    ,[file_id]
    ,[name]
    ,physical_name
    ,type_desc
    ,state_desc
    ,is_percent_growth
    ,growth
    ,CONVERT(BIGINT, growth / 128.0) AS [Growth in MB]
    ,CONVERT(BIGINT, size / 128.0) AS [Total Size in MB]
FROM sys.master_files WITH (NOLOCK)
WHERE [database_id] > 4
    AND [database_id] <> 32767
    OR [database_id] = 2
ORDER BY DB_NAME([database_id])
OPTION (RECOMPILE);

通常會查看：

數據文件是否與日志文件放到了不同的磁盤上？
難道都裝到了 C 盤上？
tempdb 是否指定了獨立的磁盤？
有幾個 tempdb 文件？
這些 tempdb 都多大了？
數據庫是否包含多個文件？
數據文件的增長步長合適嗎？

查看指定數據庫文件的大小和可用空間

需要指定數據庫，或使用 use 指定，例如 use TEST。

SELECT f.[name] AS [File Name]
    ,f.physical_name AS [Physical Name]
    ,CAST((f.size / 128.0) AS DECIMAL(15, 2)) AS [Total Size in MB]
    ,CAST(f.size / 128.0 - CAST(FILEPROPERTY(f.[name], 'SpaceUsed') AS INT) / 128.0 AS DECIMAL(15, 2)) AS [Available Space In MB]
    ,[file_id]
    ,fg.[name] AS [Filegroup Name]
FROM sys.database_files AS f WITH (NOLOCK)
LEFT OUTER JOIN sys.data_spaces AS fg WITH (NOLOCK) ON f.data_space_id = fg.data_space_id
OPTION (RECOMPILE);

服務器 Disk 容量和掛載信息

SELECT DISTINCT vs.volume_mount_point
    ,vs.file_system_type
    ,vs.logical_volume_name
    ,CONVERT(DECIMAL(18, 2), vs.total_bytes / 1073741824.0) AS [Total Size (GB)]
    ,CONVERT(DECIMAL(18, 2), vs.available_bytes / 1073741824.0) AS [Available Size (GB)]
    ,CAST(CAST(vs.available_bytes AS FLOAT) / CAST(vs.total_bytes AS FLOAT) AS DECIMAL(18, 2)) * 100 AS [Space Free %]
FROM sys.master_files AS f WITH (NOLOCK)
CROSS APPLY sys.dm_os_volume_stats(f.database_id, f.[file_id]) AS vs
OPTION (RECOMPILE);

SELECT db_name(vs.database_id) AS DatabaseName
    ,vs.file_id
    ,vs.volume_mount_point
    ,vs.volume_id
    ,vs.logical_volume_name
    ,vs.file_system_type
    ,(vs.total_bytes / 1024 / 1024 / 1024) AS [TotalSize(GB)]
    ,(vs.available_bytes / 1024 / 1024 / 1024) AS [AvailableSize(GB)]
    ,vs.supports_compression
    ,vs.supports_alternate_streams
    ,vs.supports_sparse_files
    ,vs.is_read_only
    ,vs.is_compressed
FROM sys.master_files mf
CROSS APPLY sys.dm_os_volume_stats(mf.database_id, mf.file_id) vs;

查看 Disk 剩余空間

EXEC master.dbo.xp_fixeddrives

SELECT DISTINCT SUBSTRING(volume_mount_point, 1, 1) AS Volume_mount_point
    ,total_bytes / 1024 / 1024 AS Total_MB
    ,available_bytes / 1024 / 1024 AS Available_MB
FROM sys.master_files AS f
CROSS APPLY sys.dm_os_volume_stats(f.database_id, f.file_id);

查詢數據庫設置的 Recovery Model

SELECT db.[name] AS [Database Name]
    ,db.recovery_model_desc AS [Recovery Model]
    ,db.state_desc
    ,db.log_reuse_wait_desc AS [Log Reuse Wait Description]
    ,CONVERT(DECIMAL(18, 2), ls.cntr_value / 1024.0) AS [Log Size (MB)]
    ,CONVERT(DECIMAL(18, 2), lu.cntr_value / 1024.0) AS [Log Used (MB)]
    ,CAST(CAST(lu.cntr_value AS FLOAT) / CAST(ls.cntr_value AS FLOAT) AS DECIMAL(18, 2)) * 100 AS [Log Used %]
    ,db.[compatibility_level] AS [DB Compatibility Level]
    ,db.page_verify_option_desc AS [Page Verify Option]
    ,db.is_auto_create_stats_on
    ,db.is_auto_update_stats_on
    ,db.is_auto_update_stats_async_on
    ,db.is_parameterization_forced
    ,db.snapshot_isolation_state_desc
    ,db.is_read_committed_snapshot_on
    ,db.is_auto_close_on
    ,db.is_auto_shrink_on
    ,db.target_recovery_time_in_seconds
    ,db.is_cdc_enabled
FROM sys.databases AS db WITH (NOLOCK)
INNER JOIN sys.dm_os_performance_counters AS lu WITH (NOLOCK) ON db.NAME = lu.instance_name
INNER JOIN sys.dm_os_performance_counters AS ls WITH (NOLOCK) ON db.NAME = ls.instance_name
WHERE lu.counter_name LIKE N'Log File(s) Used Size (KB)%'
    AND ls.counter_name LIKE N'Log File(s) Size (KB)%'
    AND ls.cntr_value > 0
OPTION (RECOMPILE);

通常會關注：

數據庫實例（Instance）上建立了多少個數據庫？
它們都分別使用了什么恢復模型（Recovery Model）？
Log 重用是如何設置的？
事務日志（Transaction Log）現在多大了？
兼容等級（Compatibility Level）是怎么配置的？
頁驗證選項（Page Verify Option）的設置是什么？通常為 CHECKSUM。
是否設置了 Auto Update Statistics Asynchronously 選項？
確保未開啟 auto_shrink 和 auto_close 選項。

查看最近的 Full Backup 信息

SELECT TOP (30) bs.machine_name
    ,bs.server_name
    ,bs.database_name AS [Database Name]
    ,bs.recovery_model
    ,CONVERT(BIGINT, bs.backup_size / 1048576) AS [Uncompressed Backup Size (MB)]
    ,CONVERT(BIGINT, bs.compressed_backup_size / 1048576) AS [Compressed Backup Size (MB)]
    ,CONVERT(NUMERIC(20, 2), (CONVERT(FLOAT, bs.backup_size) / CONVERT(FLOAT, bs.compressed_backup_size))) AS [Compression Ratio]
    ,DATEDIFF(SECOND, bs.backup_start_date, bs.backup_finish_date) AS [Backup Elapsed Time (sec)]
    ,bs.backup_finish_date AS [Backup Finish Date]
FROM msdb.dbo.backupset AS bs WITH (NOLOCK)
WHERE DATEDIFF(SECOND, bs.backup_start_date, bs.backup_finish_date) > 0
    AND bs.backup_size > 0
    AND bs.type = 'D' -- Change to L if you want Log backups
    AND database_name = DB_NAME(DB_ID())
ORDER BY bs.backup_finish_date DESC
OPTION (RECOMPILE);

獲取所有數據庫的 VLF 數量

VLF ：Virtual Log File

SQL Server 將日志文件 LDF 划分為多個 VLF 以提高日志處理效率。VLF 的數量和大小不能通過配置指定，SQL Server 會按情況自行判斷處理，而如果生成了過多的 VLF 則會產生性能問題。通過指定合理的日志文件初始大小和增長步長，可以有效的防止過多 VLF 的產生。

1M-64M 　　4
64M-1GB 8
>1GB 16

CREATE TABLE #VLFInfo (
    RecoveryUnitID INT
    ,FileID INT
    ,FileSize BIGINT
    ,StartOffset BIGINT
    ,FSeqNo BIGINT
    ,[Status] BIGINT
    ,Parity BIGINT
    ,CreateLSN NUMERIC(38)
    );

CREATE TABLE #VLFCountResults (
    DatabaseName SYSNAME
    ,VLFCount INT
    );

EXEC sp_MSforeachdb N'Use [?]; 

                INSERT INTO #VLFInfo 
                EXEC sp_executesql N''DBCC LOGINFO([?])''; 
     
                INSERT INTO #VLFCountResults 
                SELECT DB_NAME(), COUNT(*) 
                FROM #VLFInfo; 

                TRUNCATE TABLE #VLFInfo;'

SELECT DatabaseName
    ,VLFCount
FROM #VLFCountResults
ORDER BY VLFCount DESC;

DROP TABLE #VLFInfo;

DROP TABLE #VLFCountResults;

較高的 VLF 數量會影響寫入性能，並且會使數據庫的恢復過程變慢，通常需要保持 VLF Counts 在 200 以下。

SQL Server 的錯誤日志位置

SELECT is_enabled
    ,[path]
    ,max_size
    ,max_files
FROM sys.dm_os_server_diagnostics_log_configurations WITH (NOLOCK)
OPTION (RECOMPILE);

查詢近期的 Error Log 信息

DECLARE @Time_Start DATETIME;
DECLARE @Time_End DATETIME;

SET @Time_Start = getdate() - 2;
SET @Time_End = getdate();

-- Create the temporary table
CREATE TABLE #ErrorLog (
    logdate DATETIME
    ,processinfo VARCHAR(255)
    ,Message VARCHAR(500)
    )

-- Populate the temporary table
INSERT #ErrorLog (
    logdate
    ,processinfo
    ,Message
    )
EXEC master.dbo.xp_readerrorlog 0
    ,1
    ,NULL
    ,NULL
    ,@Time_Start
    ,@Time_End
    ,N'desc';

-- Filter the temporary table
SELECT LogDate
    ,Message
FROM #ErrorLog
WHERE (
        Message LIKE '%error%'
        OR Message LIKE '%failed%'
        )
    AND processinfo NOT LIKE 'logon'
ORDER BY logdate DESC

-- Drop the temporary table 
DROP TABLE #ErrorLog

在錯誤日志中查詢 I/O 超過 15s 的請求

CREATE TABLE #IOWarningResults (
    LogDate DATETIME
    ,ProcessInfo SYSNAME
    ,LogText NVARCHAR(1000)
    );

INSERT INTO #IOWarningResults
EXEC xp_readerrorlog 0
    ,1
    ,N'taking longer than 15 seconds';

INSERT INTO #IOWarningResults
EXEC xp_readerrorlog 1
    ,1
    ,N'taking longer than 15 seconds';

INSERT INTO #IOWarningResults
EXEC xp_readerrorlog 2
    ,1
    ,N'taking longer than 15 seconds';

INSERT INTO #IOWarningResults
EXEC xp_readerrorlog 3
    ,1
    ,N'taking longer than 15 seconds';

INSERT INTO #IOWarningResults
EXEC xp_readerrorlog 4
    ,1
    ,N'taking longer than 15 seconds';

SELECT LogDate
    ,ProcessInfo
    ,LogText
FROM #IOWarningResults
ORDER BY LogDate DESC;

DROP TABLE #IOWarningResults;

如果能夠查詢出結果，可以說明 I/O 性能存在問題，但是哪里引起的還需進一步探索。

查詢 Disk 的性能指標

SELECT [Drive]
    ,CASE 
        WHEN num_of_reads = 0
            THEN 0
        ELSE (io_stall_read_ms / num_of_reads)
        END AS [Read Latency (ms)]
    ,CASE 
        WHEN io_stall_write_ms = 0
            THEN 0
        ELSE (io_stall_write_ms / num_of_writes)
        END AS [Write Latency (ms)]
    ,CASE 
        WHEN (
                num_of_reads = 0
                AND num_of_writes = 0
                )
            THEN 0
        ELSE (io_stall / (num_of_reads + num_of_writes))
        END AS [Overall Latency (ms)]
    ,CASE 
        WHEN num_of_reads = 0
            THEN 0
        ELSE (num_of_bytes_read / num_of_reads)
        END AS [Avg Bytes/Read]
    ,CASE 
        WHEN io_stall_write_ms = 0
            THEN 0
        ELSE (num_of_bytes_written / num_of_writes)
        END AS [Avg Bytes/Write]
    ,CASE 
        WHEN (
                num_of_reads = 0
                AND num_of_writes = 0
                )
            THEN 0
        ELSE ((num_of_bytes_read + num_of_bytes_written) / (num_of_reads + num_of_writes))
        END AS [Avg Bytes/Transfer]
FROM (
    SELECT LEFT(UPPER(mf.physical_name), 2) AS Drive
        ,SUM(num_of_reads) AS num_of_reads
        ,SUM(io_stall_read_ms) AS io_stall_read_ms
        ,SUM(num_of_writes) AS num_of_writes
        ,SUM(io_stall_write_ms) AS io_stall_write_ms
        ,SUM(num_of_bytes_read) AS num_of_bytes_read
        ,SUM(num_of_bytes_written) AS num_of_bytes_written
        ,SUM(io_stall) AS io_stall
    FROM sys.dm_io_virtual_file_stats(NULL, NULL) AS vfs
    INNER JOIN sys.master_files AS mf WITH (NOLOCK) ON vfs.database_id = mf.database_id
        AND vfs.file_id = mf.file_id
    GROUP BY LEFT(UPPER(mf.physical_name), 2)
    ) AS tab
ORDER BY [Overall Latency (ms)]
OPTION (RECOMPILE);

通常 Latency 的值大於 20-25 ms 時可考慮有性能問題。

查看哪個數據庫文件 I/O 瓶頸最嚴重

SELECT DB_NAME(fs.database_id) AS [Database Name]
    ,CAST(fs.io_stall_read_ms / (1.0 + fs.num_of_reads) AS NUMERIC(10, 1)) AS [avg_read_stall_ms]
    ,CAST(fs.io_stall_write_ms / (1.0 + fs.num_of_writes) AS NUMERIC(10, 1)) AS [avg_write_stall_ms]
    ,CAST((fs.io_stall_read_ms + fs.io_stall_write_ms) / (1.0 + fs.num_of_reads + fs.num_of_writes) AS NUMERIC(10, 1)) AS [avg_io_stall_ms]
    ,CONVERT(DECIMAL(18, 2), mf.size / 128.0) AS [File Size (MB)]
    ,mf.physical_name
    ,mf.type_desc
    ,fs.io_stall_read_ms
    ,fs.num_of_reads
    ,fs.io_stall_write_ms
    ,fs.num_of_writes
    ,fs.io_stall_read_ms + fs.io_stall_write_ms AS [io_stalls]
    ,fs.num_of_reads + fs.num_of_writes AS [total_io]
FROM sys.dm_io_virtual_file_stats(NULL, NULL) AS fs
INNER JOIN sys.master_files AS mf WITH (NOLOCK) ON fs.database_id = mf.database_id
    AND fs.[file_id] = mf.[file_id]
ORDER BY avg_io_stall_ms DESC
OPTION (RECOMPILE);

考慮將數據庫文件移動到不同的磁盤上，或更快的磁盤陣列上以改進性能。

按照 Write I/O 進行排名

SELECT [ReadLatency] = CASE 
        WHEN [num_of_reads] = 0
            THEN 0
        ELSE ([io_stall_read_ms] / [num_of_reads])
        END
    ,[WriteLatency] = CASE 
        WHEN [num_of_writes] = 0
            THEN 0
        ELSE ([io_stall_write_ms] / [num_of_writes])
        END
    ,[Latency] = CASE 
        WHEN (
                [num_of_reads] = 0
                AND [num_of_writes] = 0
                )
            THEN 0
        ELSE ([io_stall] / ([num_of_reads] + [num_of_writes]))
        END
    ,[AvgBytesPerRead] = CASE 
        WHEN [num_of_reads] = 0
            THEN 0
        ELSE ([num_of_bytes_read] / [num_of_reads])
        END
    ,[AvgBytesPerWrite] = CASE 
        WHEN [num_of_writes] = 0
            THEN 0
        ELSE ([num_of_bytes_written] / [num_of_writes])
        END
    ,[AvgBytesPerTransfer] = CASE 
        WHEN (
                [num_of_reads] = 0
                AND [num_of_writes] = 0
                )
            THEN 0
        ELSE (([num_of_bytes_read] + [num_of_bytes_written]) / ([num_of_reads] + [num_of_writes]))
        END
    ,LEFT([mf].[physical_name], 2) AS [Drive]
    ,DB_NAME([vfs].[database_id]) AS [DB]
    ,[mf].[physical_name]
    ,[mf].file_id
FROM sys.dm_io_virtual_file_stats(NULL, NULL) AS [vfs]
JOIN sys.master_files AS [mf] ON [vfs].[database_id] = [mf].[database_id]
    AND [vfs].[file_id] = [mf].[file_id]
ORDER BY [WriteLatency] DESC;

獲取數據庫的 I/O 使用率

WITH Aggregate_IO_Statistics
AS (
    SELECT DB_NAME(database_id) AS [Database Name]
        ,CAST(SUM(num_of_bytes_read + num_of_bytes_written) / 1048576 AS DECIMAL(12, 2)) AS io_in_mb
    FROM sys.dm_io_virtual_file_stats(NULL, NULL) AS [DM_IO_STATS]
    GROUP BY database_id
    )
SELECT ROW_NUMBER() OVER (
        ORDER BY io_in_mb DESC
        ) AS [I/O Rank]
    ,[Database Name]
    ,io_in_mb AS [Total I/O (MB)]
    ,CAST(io_in_mb / SUM(io_in_mb) OVER () * 100.0 AS DECIMAL(5, 2)) AS [I/O Percent]
FROM Aggregate_IO_Statistics
ORDER BY [I/O Rank]
OPTION (RECOMPILE);

查看指定數據庫文件的 I/O 狀況

需要指定數據庫，或使用 use 指定，例如 use TEST。

SELECT DB_NAME(DB_ID()) AS [Database Name]
    ,df.[name] AS [Logical Name]
    ,vfs.[file_id]
    ,df.physical_name AS [Physical Name]
    ,vfs.num_of_reads
    ,vfs.num_of_writes
    ,vfs.io_stall_read_ms
    ,vfs.io_stall_write_ms
    ,CAST(100. * vfs.io_stall_read_ms / (vfs.io_stall_read_ms + vfs.io_stall_write_ms) AS DECIMAL(10, 1)) AS [IO Stall Reads Pct]
    ,CAST(100. * vfs.io_stall_write_ms / (vfs.io_stall_write_ms + vfs.io_stall_read_ms) AS DECIMAL(10, 1)) AS [IO Stall Writes Pct]
    ,(vfs.num_of_reads + vfs.num_of_writes) AS [Writes + Reads]
    ,CAST(vfs.num_of_bytes_read / 1048576.0 AS DECIMAL(10, 2)) AS [MB Read]
    ,CAST(vfs.num_of_bytes_written / 1048576.0 AS DECIMAL(10, 2)) AS [MB Written]
    ,CAST(100. * vfs.num_of_reads / (vfs.num_of_reads + vfs.num_of_writes) AS DECIMAL(10, 1)) AS [# Reads Pct]
    ,CAST(100. * vfs.num_of_writes / (vfs.num_of_reads + vfs.num_of_writes) AS DECIMAL(10, 1)) AS [# Write Pct]
    ,CAST(100. * vfs.num_of_bytes_read / (vfs.num_of_bytes_read + vfs.num_of_bytes_written) AS DECIMAL(10, 1)) AS [Read Bytes Pct]
    ,CAST(100. * vfs.num_of_bytes_written / (vfs.num_of_bytes_read + vfs.num_of_bytes_written) AS DECIMAL(10, 1)) AS [Written Bytes Pct]
FROM sys.dm_io_virtual_file_stats(DB_ID(), NULL) AS vfs
INNER JOIN sys.database_files AS df WITH (NOLOCK) ON vfs.[file_id] = df.[file_id]
OPTION (RECOMPILE);

協助從 I/O 角度來觀察數據庫文件所承載的壓力。

找出 I/O 平均使用最多的語句

SELECT TOP (50) OBJECT_NAME(qt.objectid, dbid) AS [SP Name]
    ,(qs.total_logical_reads + qs.total_logical_writes) / qs.execution_count AS [Avg IO]
    ,qs.execution_count AS [Execution Count]
    ,SUBSTRING(qt.[text], qs.statement_start_offset / 2 + 1, (
            CASE 
                WHEN qs.statement_end_offset = - 1
                    THEN LEN(CONVERT(NVARCHAR(max), qt.[text])) * 2
                ELSE qs.statement_end_offset
                END - qs.statement_start_offset
            ) / 2) AS [Query Text]
FROM sys.dm_exec_query_stats AS qs WITH (NOLOCK)
CROSS APPLY sys.dm_exec_sql_text(qs.sql_handle) AS qt
WHERE qt.[dbid] = DB_ID()
ORDER BY [Avg IO] DESC
OPTION (RECOMPILE);

參考資料：

How To: Troubleshooting SQL Server I/O bottlenecks

查詢正在等待 I/O 的請求等待時間

SELECT DB_NAME(database_id) AS [DBNAME]
    ,file_id
    ,io_stall
    ,io_pending_ms_ticks
    ,scheduler_address
FROM sys.dm_io_virtual_file_stats(NULL, NULL) iovfs
    ,sys.dm_io_pending_io_requests AS iopior
WHERE iovfs.file_handle = iopior.io_handle

《人人都是 DBA》系列文章索引：

序號	名稱
1	人人都是 DBA（I）SQL Server 體系結構
2	人人都是 DBA（II）SQL Server 元數據
3	人人都是 DBA（III）SQL Server 調度器
4	人人都是 DBA（IV）SQL Server 內存管理
5	人人都是 DBA（V）SQL Server 數據庫文件
6	人人都是 DBA（VI）SQL Server 事務日志
7	人人都是 DBA（VII）B 樹和 B+ 樹
8	人人都是 DBA（VIII）SQL Server 頁存儲結構
9	人人都是 DBA（IX）服務器信息收集腳本匯編
10	人人都是 DBA（X）資源信息收集腳本匯編
11	人人都是 DBA（XI）I/O 信息收集腳本匯編
12	人人都是 DBA（XII）查詢信息收集腳本匯編
13	人人都是 DBA（XIII）索引信息收集腳本匯編
14	人人都是 DBA（XIV）存儲過程信息收集腳本匯編
15	人人都是 DBA（XV）鎖信息收集腳本匯編

本系列文章《人人都是 DBA》由 Dennis Gao 發表自博客園，未經作者本人同意禁止任何形式的轉載，任何自動或人為的爬蟲轉載行為均為耍流氓。

免責聲明！

本站轉載的文章為個人學習借鑒使用，本站對版權不負任何法律責任。如果侵犯了您的隱私權益，請聯系本站郵箱yoyou2525@163.com刪除。

猜您在找 人人都是 DBA（XIV）存儲過程信息收集腳本匯編人人都是 DBA（XV）鎖信息收集腳本匯編人人都是 DBA（X）資源信息收集腳本匯編人人都是 DBA（XII）查詢信息收集腳本匯編人人都是 DBA（I）SQL Server 體系結構 powersploit的兩個信息收集的腳本 IDAPython腳本之收集函數的調用信息騰訊SRC挖掘信息收集小腳本 I/O iostat 用於輸出 CPU 和磁盤 I/O 相關的統計信息。