Microsoft SQL Server 2008 R2 Unleashed- P234 pptx

ptg
2294
CHAPTER 55 Configuring, Tuning, and Optimizing SQL Server Options
then explicitly allow cross-database ownership chaining using the
SET clause of the ALTER
DATABASE
statement.
cursor threshold
Type: Advanced
Default value:
-1
The cursor threshold option allows you to specify when SQL Server should generate a
cursor result set asynchronously. If the optimizer estimates that the number of rows
returned by the cursor is greater than the value specified by this parameter, it generates
the result set asynchronously. The optimizer makes this decision based on the distribution
statistics for each table that is participating in the join in the cursor.
To determine the optimal value for this parameter, you need to make sure that statistics
are up-to-date (by running update statistics) for the tables used in the cursors. By default,
SQL Server generates a cursor result set synchronously. If you are using a fair number of
cursors that return a large number of result sets, setting this value to a higher value results
in better performance. Setting this value to
0 forces SQL Server to always generate a cursor
result set asynchronously.
The following is an example of this option:
exec sp_configure ‘cursor threshold’, 100000
go
RECONFIGURE
go
The following is a general recommendation of what to specify based on the different appli-
cation types you are running:
. OLTP—Use the default value,

-1.
. Data warehouse—A data warehousing environment is the largest potential user of
this option due to the high volume of result rows returned by applications using
data warehouses. Setting this value to
100,000 is a good starting point.
. OLAP—Use the default value,
-1.
. Mixed—Use the default value,
-1.
default full-text language
Type: Advanced
Default value:
1033
The default full-text language option is used to specify a default language value for
full-text indexed columns. Linguistic analysis is performed on all data that is full-text
indexed and is strictly dependent on the language of the data. The default value of this
option is the language of the server.
ptg
2295
Configuration Options and Performance
55
The value of the default full-text language option is used when no language is speci-
fied for a column through the
LANGUAGE option in the CREATE FULLTEXT INDEX or ALTER
FULLTEXT INDEX
statements. If the default full-text language option is not supported
or the linguistic analysis package is not available, the
CREATE or ALTER operation doesn’t
work, and SQL Server returns an
invalid language specified error message. To see the

list of linguistic analysis packages that are part of SQL Server 2008, you query the system
view
sys.fulltext_languages, as in this example:
SELECT * FROM sys.fulltext_languages
order by 2
Go
lcid name

1025 Arabic
1093 Bengali (India)
1046 Brazilian
2057 British English
1026 Bulgarian
1027 Catalan
3076 Chinese (Hong Kong SAR, PRC)
5124 Chinese (Macau SAR)
4100 Chinese (Singapore)
1050 Croatian
1043 Dutch
1033 English
1036 French
1031 German
1095 Gujarati
1037 Hebrew
1081 Hindi
1039 Icelandic
1057 Indonesian
1040 Italian
1041 Japanese
1099 Kannada

1042 Korean
1062 Latvian
1063 Lithuanian
1086 Malay - Malaysia
1100 Malayalam
1102 Marathi
0 Neutral
1044 Norwegian (Bokmål)
2070 Portuguese
1094 Punjabi
1048 Romanian
ptg
2296
CHAPTER 55 Configuring, Tuning, and Optimizing SQL Server Options
1049 Russian
3098 Serbian (Cyrillic)
2074 Serbian (Latin)
2052 Simplified Chinese
1051 Slovak
1060 Slovenian
3082 Spanish
1053 Swedish
1097 Tamil
1098 Telugu
1054 Thai
1028 Traditional Chinese
1058 Ukrainian
1056 Urdu
1066 Vietnamese
If setting to other than the default, the sp_configure command would look like this

(when set to French):
exec sp_configure ‘default full-text language’, 1036
go
RECONFIGURE
Go
default language
Type: Basic
Default value:
0
The default language option specifies the language ID currently in use by SQL Server.
The default value is
0, which specifies the U.S. English system. As you add languages on
the server, SQL Server assigns a new ID for each language. You can then use these IDs to
specify the default language of your choice. You can add languages by using the SQL
Server Setup program. Adding a language allows SQL Server to display error messages and
date/time values in the format that is appropriate for that language. You set this option in
the Server Settings tab of the SQL Server Properties dialog.
You can override the default language for a login by using
sp_addlogin or
sp_defaultlanguage. The default language for a session is the language for that session’s
login, unless overridden on a per-session basis by using the ODBC or OLE DB APIs. To see
the languages supported for your SQL Server instance, you use the system view
sys.syslanguages.
SELECT langid, substring(name,1,20) AS name,
substring(alias,1,20) AS alias,
lcid
FROM sys.syslanguages
ptg
2297
Configuration Options and Performance

55
order by 3
Go
langid name alias lcid

31 Arabic Arabic 1025
27 Português (Brasil) Brazilian 1046
23 British British English 2057
20 Bulgarian 1026
16 hrvatski Croatian 1050
12 Czech 1029
4 Dansk Danish 1030
7 Nederlands Dutch 1043
0 us_english English 1033
24 eesti Estonian 1061
10 Suomi Finnish 1035
2 Français French 1036
1 Deutsch German 1031
19 Greek 1032
13 magyar Hungarian 1038
6 Italiano Italian 1040
3 Japanese 1041
29 Korean 1042
25 Latvian 1062
26 Lithuanian 1063
8 Norsk Norwegian 2068
14 polski Polish 1045
9 Português Portuguese 2070
15 românä Romanian 1048
21 Russian 1049

30 Simplified Chinese 2052
17 Slovak 1051
18 slovenski Slovenian 1060
5 Español Spanish 3082
11 Svenska Swedish 1053
32 Thai 1054
28 Traditional Chinese 1028
22 Türkçe Turkish 1055
Setting the default language to something other than the default would look like this
(again, when set to French):
exec sp_configure ‘default language’, 2
go
ptg
2298
CHAPTER 55 Configuring, Tuning, and Optimizing SQL Server Options
FIGURE 55.4 SQL Server 2008 language configuration from SSMS.
RECONFIGURE
Go
You can change the language for a session during the session through the SET LANGUAGE
statement. From SSMS, you can easily see the current default language and default full-text
language settings, as shown in Figure 55.4.
EKM provider enabled
Type: basic
Default value: 0
The EKM provider enabled option controls Extensible Key Management (EKM) device
support in SQL Server. By default, this option is off. SQL Server 2008 Extensible Key
Management enables third-party EKM/HSM vendors to register their modules in SQL
Server. When registered, SQL Server 2008 users can use the encryption keys stored on EKM
modules. This enables SQL Server to access the advanced encryption features these
modules support, such as bulk encryption and decryption, and key management functions

such as key aging and key rotation.
ptg
2299
Configuration Options and Performance
55
The following is an example of this option:
exec sp_configure ‘EKM provider enabled’, 1
go
RECONFIGURE
go
filestream_access_level
Type: basic
Default value:
0
You use the filestream_access_level option to change the FILESTREAM access level.
Before this option has any effect, the Windows administration settings for FILESTREAM
must be enabled. You can also enable these settings when you are installing SQL Server or
by using SQL Server Configuration Manager (SSCM) for a specific SQL Server instance.
Setting the value to
0 disables this option. A value of 1 enables FILESTREAM for T-SQL
access only. A value of
2 enables FILESTREAM for both T-SQL and Win32 streaming access.
FILESTREAM integrates the SQL Server Database Engine with an NTFS file system by
storing
varbinary(max) binary large object (BLOB) data as files on the file system. T-SQL
statements can insert, update, query, search, and back up FILESTREAM data.
The following is an example of this option:
exec sp_configure ‘filestream_access_level’, 2
go
RECONFIGURE

fill factor
Type: Basic
Default value: 0
The fill factor option allows you to define the percentage of free space on a data page
or an index page when you create an index or a table. The value can range from
1 to 100.
Setting the value to
80 would mean each page would be 80% full at the time of the create
index
. SQL Server also allows you to specify the value of fill factor at the server level by
providing a
fill factor parameter.
The following is an example of this option:
exec sp_configure ‘fill factor’, 80
go
RECONFIGURE
Go
ptg
2300
CHAPTER 55 Configuring, Tuning, and Optimizing SQL Server Options
The following is a general recommendation of what to specify based on the different
application types you are running:
. OLTP—This is a good candidate for leaving space free in pages due to the update,
delete, and insert characteristics. Try 80% full value first and watch the page split
activity.
. Data warehouse—Use the default value,
0.
. OLAP—Use the default value,
0.
. Mixed—Use the default value, 0, or about a 90% full value.

index create memory
Type: Advanced, self-configuring
Default value: 0
The index create memory option is used to control the amount of memory used by index
creation sorts. It is a self-configuring option and usually doesn’t need to be adjusted.
However, if you are having problems with the creation of large indexes, you might want
to try specifying a KB value here that will contain the sort portion of the
index create.
The following is an example of this option:
exec sp_configure ‘index create memory’, 1000
go
RECONFIGURE
go
in-doubt xact resolution
Type: Advanced
Default value:
0
The in-doubt xact resolution option changes the behavior of the default outcome for
transactions that the Microsoft Distributed Transaction Coordinator (MS DTC) is unable to
resolve. This helps prevent MS DTC–related downtime and provides clarity to transaction
outcomes across database servers. A value of
1 is a presume commit directive. In other
words, any MS DTC in-doubt transactions are presumed to have committed. A value of 2
is a presume abort directive, which means any MS DTC in-doubt transactions are presumed
to have aborted.
The following is an example of this option:
exec sp_configure ‘in-doubt xact resolution’, 2
go
RECONFIGURE
go

ptg
2301
Configuration Options and Performance
55
lightweight pooling
Type: Advanced
Default value: 0
Lightweight pooling is relevant to multiprocessor environments that have excessive
context switching. By flipping the
lightweight pooling switch, you might get better
throughput by performing the context switching inline, thus helping to reduce
user/kernel ring transitions. The
lightweight pooling option actually causes SQL Server
to switch to fiber mode scheduling. CLR execution is not supported under lightweight
pooling. You can disable one of two options:
clr enabled or lightweight pooling.
The following is an example of this option:
exec sp_configure ‘lightweight pooling’, 1
go
RECONFIGURE
go
The following is a general recommendation of what to specify based on the different
application types you are running:
. OLTP—This is a good candidate for use on a multiprocessor machine.
. Data warehouse—This has a good potential for usage on a multiprocessor machine.
. OLAP—Use the default value,
0.
. Mixed—Use the default value,
0.
locks

Type: Advanced, self-configuring
Default value:
0
In versions of SQL Server prior to 2005, the DBA had to specify the number of locks avail-
able to SQL Server. If this parameter was set to a low value, a query requiring a large
number of locks would fail at runtime. Setting it too high would result in wasting memory
that otherwise could be used to cache data. SQL Server 2008 can handle locks dynamically
if this parameter is set to the default value (
0). SQL Server initially allocates 2% of memory
available to SQL Server (or usually about an initial pool of 2,500 lock structures). As lock
resource structures are consumed, the lock manager allocates more lock resources to the
pool, to a maximum of 60% of the memory available on SQL Server. When the locks
option is not set to
0, lock escalation occurs when the number of locks reaches 40% of the
value specified for locks. Unless you are certain of the overall lock consumption of your
application, you probably don’t need to change this value.
ptg
2302
CHAPTER 55 Configuring, Tuning, and Optimizing SQL Server Options
The following is an example of this option:
sp_configure ‘locks’, 10000
go
RECONFIGURE
go
max degree of parallelism
Type: Advanced
Default value:
0
The max degree of parallelism option specifies the number of threads to be used for
parallel query execution. On a single-processor server, this value is always ignored. For

multiprocessor servers, a default value of
0 signifies that all the CPUs will be used for
parallel query execution. When SQL Server 2008 runs on a machine that has more than
one processor, it detects the best degree of parallelism. If you set this value to
1, all query
plans are serialized. If the
affinity mask option is on, parallel query execution takes place
only on the CPUs for which the
affinity mask bit is turned on. In that way, these two
options can be used in conjunction. The application types assessment is the same as
described for the
affinity mask option.
The following is an example of this option:
sp_configure ‘max degree of parallelism’, 4
go
RECONFIGURE
go
max server memory and min server memory
Type: Advanced, self-configuring
Default values:
2147483647 and 0
The max server memory option specifies the maximum amount of memory (in megabytes)
available to SQL Server. It is used in conjunction with
min server memory, and they essen-
tially establish upper and lower bounds for memory allocation. SQL Server uses this
memory for user connections, locks, internal data structures, and caching of the data. This
is the memory pool described earlier. The default value of
2147483647 for the max server
memory
option means that SQL Server performs dynamic allocation of memory from the

operating system, based on available physical memory on the machine. The default value
of
0 for the min server memory option means that SQL Server starts allocation memory as
it is needed and then never goes below the minimum value after it is reached.
ptg
2303
Configuration Options and Performance
55
The SQL Server lazywriter process is responsible for making sure enough memory is avail-
able to SQL Server for the optimal number of buffers and Windows so that no excess
paging occurs at the operating system level. The lazywriter process frequently checks phys-
ical memory available on the machine. If the memory available is greater than 5MB, lazy-
writer assigns excess memory to the SQL Server buffer cache.
You should watch the
Working Set performance counter, which shows the amount of
memory used by a process (SQL Server in this case). If this number is consistently below
the amount of memory for which SQL Server is configured, SQL Server is configured for
more memory than it needs. You can also adjust the
set working set size configura-
tion option.
If SQL Server is the only application running on a machine, you might want to perform
static memory allocation. You need to be careful when you allocate fixed memory to SQL
Server. If you allocate more memory to SQL Server than the machine has, SQL Server fails
to start. You use the
-f option during startup to bring up SQL Server with the default
configuration. You can then change the value to the correct value and restart SQL Server.
The following is an example of this option:
exec sp_configure ‘max server memory’, 200
go
RECONFIGURE

Go
exec sp_configure ‘min server memory’, 8
go
RECONFIGURE
go
Because memory is managed dynamically, these settings basically create the lower and
upper bounds within which SQL Server must operate. When SQL Server is started, it uses
the minimum server memory amount to pre-allocate space to SQL Server. If the load on
the server never requires allocating the amount of memory specified in the
minimum
server memory
option, SQL Server runs with less memory.
Figure 55.5 shows both the
minimum server memory and maximum server memory settings
on the Server Properties page.
The following is a general recommendation of what to specify based on the different
application types you are running:
. OLTP—For those with heavy loads, this is a good candidate for high minimum
memory settings.
. Data warehouse—Use the default values.
. OLAP—Use the default values.
. Mixed—For those with heavy loads, this is a good candidate for high minimum
memory settings.