Set-ItemProperty Registry::HKEY_CLASSES_ROOT\.csvh -Name PerceivedType -Value text

• Path – this is the l ocation of the file that we want to work with
• Literalpath – this is used in the place of Path if we need to work with special characters (like wildcards) in our path string. No interpretation of the wildcards is done when passed in via the LiteralPath parameter
• Algorithm – this is the type of hash we want to create. For our purposes we are not trying to use this for encryption purposes so MD5 and SHA1 are as valid as SHA256 or any other available options. By default SHA256 is used.
• InputStream – this is used if the file is opened as a stream rather than directly from the file using the -Path parameter

Get-FileHash -Path D:\ISO\en_sql_server_2016_service_pack_1_x64_dvd_9542248.iso | Format-List

Get-ChildItem D:\HashTest | Get-FileHash

$IsoHash = Get-ChildItem D:\HashTest | Get-FileHash -Algorithm MD5
$IsoHash[2].Hash
33B324D311FDFC68A9A3E3405B8CD222
$IsoHash[2].Path
D:\ISO\en_sql_server_2016_service_pack_1_x64_dvd_9542248.iso
$IsoHash[2].Algorithm
MD5

$FileToConfirm = "en_sql_server_2016"
$r = $IsoHash.path -match $FileToConfirm
if ($r) {
"$r found" | Write-Output
}
else {
"No details found for $FileToConfirm" | Write-Output
}
D:\HashTest\en_sql_server_2016_service_pack_1_x64_dvd_9542248.iso found

# export to a file
$IsoHash | Export-csv -Path "D:\HashTest\ISOArchive.csvh"

# read hash from file and compare with newly calculated hash
$ISOArchive = @()
$ISOArchive = Import-csv -path "D:\hashtest\ISOArchive.csvh"

# choose the file we want to verify and get the hash of it right now
$FileToCheck = "D:\hashtest\en_sql_server_2016_service_pack_1_x64_dvd_9542248.iso"
$NewFileHash = Get-FileHash $FileToCheck -Algorithm MD5

# check if the new hash is found in the archive
$match = ($ISOArchive.hash) -match [regex]::Escape($NewFileHash.Hash)
if ($match) {
    "The hash for $($match.path) matches the archive value. It is safe to use." | Write-Output
}
else { # check if the file is found in the archive
    $exists = (($ISOArchive.path) -match [regex]::Escape($NewFileHash.Path)) -and (($ISOArchive.algorithm) -match [regex]::Escape($NewFileHash.Algorithm))
    if ($exists) { # if the file exists and the hash wasn’t found then the file has changed
        "The hash for $($match.path) does not match the archive value. It is NOT safe." | Write-Warning
    }
    else { # if the file isn’t found then we don’t have a hash that we can compare with
        if (!((($ISOArchive.path) -match [regex]::Escape($NewFileHash.Path)))) {
            "It doesnt look like there is a hash value for $($Newfilehash.path) in the archive." | Write-Output
        }
        else { # we have to compare hashes generated with the same algorithm
            "Please re-test $($newfilehash.path) using the correct hash algorithm" | Write-Output
        }
    }
}

[System.Io.Path]::GetTempFileName()

C:\Users\username\AppData\Local\Temp\tmpE412.tmp

$F = [System.Io.Path]::GetTempFileName() 

Get-Process | Out-File $F -Append

"Process information stored in $F" | Write-Output

$F = [System.Io.Path]::GetTempFileName() 
$F | Remove-Item -Force

Get-Process | Out-File $F

"Process information stored in $F" | Write-Output

$SMOServer = New-Object ('Microsoft.SqlServer.Management.Smo.Server') "$env:computername\sql2016"

$LMode = $SMOServer.LoginMode

$LMode.GetType()

IsPublic IsSerial Name                                     BaseType
-------- -------- ----                                     --------
True     True     ServerLoginMode                          System.Enum

[System.Enum]::GetValues([Microsoft.SqlServer.Management.SMO.ServerLoginMode])

Normal 
Integrated 
Mixed 
Unknown

Get-Service | ? Name -match mssql\$

Status   Name               DisplayName
------   ----               -----------
Stopped  MSSQL$SQL2008A     SQL Server (SQL2008A)
Stopped  MSSQL$SQL2008B     SQL Server (SQL2008B)
Running  MSSQL$SQL2014      SQL Server (SQL2014)
Running  MSSQL$SQL2016      SQL Server (SQL2016)

$svc = Get-Service | ? Name -match mssql\$ | select -First 1

$svc_status = $svc.Status

$svc_status.GetType()

IsPublic IsSerial Name                                     BaseType
-------- -------- ----                                     --------
True     True     ServiceControllerStatus                  System.Enum

Unable to find type [ServiceControllerStatus].

$svc_status.GetType() | select *

$svc_status.GetType() | select fullname, namespace

FullName                                      Namespace
--------                                      ---------
System.ServiceProcess.ServiceControllerStatus System.ServiceProcess

[System.Enum]::GetValues([system.ServiceProcess.ServiceControllerStatus])

Stopped 
StartPending
StopPending
Running
ContinuePending
PausePending
Paused

1. Add a Clustered Index

2. Add a Fill Factor

3. Add Row Compression (the specified fill factor will persist through the rebuild)

4. Add Page Compression (includes Row Compression which will be applied first)

Working at a customer site, and I happened to notice that the Exception ring buffer contained a lot (and I mean A.L.O.T.) of errors. Ring Buffers were introduced in SQL 2008, and contain a fixed size cycle of records for the ring buffer they pertain to. Now… note – sys.dm_os_ring_buffers is a DMV which is NOT SUPPORTED. and future compatibility is NOT guaranteed… But… that doesn’t mean we can’t use them – right …

So the ring buffers available are :-

[sql]SELECT DISTINCT ring_buffer_type FROM sys.dm_os_ring_buffers[/sql]

Which gives me:

For this Blog Post, I am particularly interested in the RING_BUFFER_EXCEPTION ring buffer, as this will show me the exceptions which have been occurring across the instance. Since these exceptions might only be a severity of 15 or below, it is entirely possible that we don’t know they are occurring – unless we look in the ring buffer.

But how?

If I cast the ring buffer record as an XML datatype, I can see the format of the payload:

So, I really need to break this XML down so I can do something useful with it, and whilst I am at it, I can use the sys.messages table to turn the Error (<Error>926</Error> in this case) into something slightly more meaningful, with:

[sql]SELECT<br> DATEADD(ms, -1 * (info.ms_ticks - XEvent.value('(@time)[1]', 'INT')), GETDATE()) AS [TimeStamp]<br> , XEvent.value('(@id)[1]', 'INT') AS [ID]<br> , XEvent.value('(./Exception/Error)[1]', 'INT') AS [Error]<br> , XEvent.value('(./Exception/Severity)[1]', 'INT') AS [Severity]<br> , XEvent.value('(./Exception/State)[1]', 'INT') AS [State]<br> , XEvent.value('(./Exception/UserDefined)[1]', 'INT') AS [UserDefined]<br> , sm.text<br> FROM (<br> SELECT<br> CAST(record AS XML) AS [target_data]<br> FROM<br> sys.dm_os_ring_buffers<br> WHERE<br> ring_buffer_type = 'RING_BUFFER_EXCEPTION' ) as XEventsRingBuffer<br> CROSS APPLY target_Data.nodes('/Record') AS XEventData (XEvent)<br> INNER JOIN sys.messages sm<br> ON sm.message_id = XEvent.value('(./Exception/Error)[1]', 'INT')<br> CROSS APPLY sys.dm_os_sys_info info[/sql]

Resulting in:

Ok – that’s cool. So I can now use the ring buffer to extract the XML of the exception and turn it into something a bit more human readable.

What if I need to investigate further? What if the Exception leads me to something not so straight-forward? Well, XEvents to the rescue:

[sql]-- If the Event Sessions already Exists, drop it<br> IF EXISTS ( SELECT 1 FROM [sys].[server_event_sessions] WHERE [name] = N'XE_Error_Capture')<br> BEGIN<br> DROP EVENT SESSION [XE_Error_Capture] ON SERVER;<br> END<br> -- Create the XEvent to catch the error<br> CREATE EVENT SESSION [XE_Error_Capture]<br> ON SERVER<br> ADD EVENT [sqlserver].[error_reported] (<br> ACTION([sqlserver].[session_id],<br> [sqlserver].[database_name],<br> [sqlserver].[tsql_stack])<br> WHERE ([error_number]=123456789)) -- Replace with Error Number&nbsp;&nbsp; <br> ADD TARGET [package0].[event_file](SET FILENAME = N'C:\Temp\XE_Error_Capture.xel')<br> GO<br> -- Start the Session<br> ALTER EVENT SESSION [XE_Error_Capture] ON SERVER<br> STATE = START;<br> GO[/sql]

I can now use an XEvent to specifically hunt out the Errors previously discovered in the ring buffer – including the database which they are occurring in, and the tsql_stack (should I need it). This can be broken down in the usual manner, or even viewed directly with the XEvent Viewer built right into SSMS.

Neat

• What they are
• Where (and why) they’re useful
• How to create one
• How to convince the optimizer to use them

What is a filtered index?

When might you want to use one?

• When indexing a column with a lot of NULL values.  Here’s an example from AdventureWorks:

• When you want to facilitate access to subsets of data:

• Of course, useful indexes are seldom defined against a single column.  Perhaps you need to “cover” different sets of columns depending on which Status you’re querying for:

• You might also want to filter on various date ranges, but know that this is a bit trickier to implement since you can’t do something like the highlighted code:

So, how do you create one?

• You can’t create a filtered index against a view (although a view can leverage filtered index defined on a table)
• Errors will occur if column data is implicitly or explicitly converted
• You cannot reference a computed column, a UDT column, a spatial or hierarchyID data type column
• Filtered indexes do not apply to XML indexes and full-text indexes

Why won’t the Optimizer use my filtered index?

Coercing the Optimizer to do our bidding

• Add literal values that align with your filtered index.  These will have to updated periodically if you modify your filtered indexes – say to follow a date range.
• Query a view that’s based on literal values that align with your filtered index then query the view instead of the base table.   You can update the view periodically to cover a new date range that also aligns with a filtered index.
• Use a RECOMPILE hint – this prevents the plan from being cached, reuse isn’t an issue, so the Optimizer is OK with using the filtered index.