22nd International Unicode Conference San Jose, CA September 20021 Going Global With SQL Server 2000 Beverly Sherry Program Manager Global Release Services.

22nd International Unicode Conference San Jose, CA September 2002 1

Going Global With SQL Server 2000 Going Global With SQL Server 2000

Beverly Sherry

Program Manager

Global Release Services for SQL Server

Microsoft Corporation



SQL Server Setup

• Clean Install/Collation

• Upgrades/Collation


Collation

• Collation what is it and why do we need it to support our multilingual data?


Collation in SQL Server 6.5 and earlier versions

• No Unicode support

• One code page per server

• One collation per server


Collation in SQL Server 7.0

• Unicode datatypes supported

• Two collations– Unicode– Non-Unicode

• Number of collations distilled down to the minimum necessary

• Collation consistency across OS’s


Collation in SQL Server 2000

• Combined code pages and Unicode collations into a single entity

• Flexible model to specify collations at a more granular level


"Windows" collations

• 43 language collations– Added for unique code pages

– Added for unique ordering

• Suffix meanings– _BIN (Binary)

– _CI/_CS (Case sensitivity)

– _AI/_AS (Accent sensitivity)

– _KS - kanatype sensitivity (hiragana/katakana)

– _WS - width sensitivity (full/half width)


SQL Collations

• Provided for backwards compatibility with prior versions of SQL Server


Collation at four levels

• Server

• Database

• Column

• Expression


Collation at the server level

• Acts as a default for all databases

• Can be changed with RebuildM.exe in the tools\BINN dir – why do you want to do this and how.

• Querying the server collation:

SELECT CONVERT(char, SERVERPROPERTY('collation'))


Upgrade Path

• Unicode datatypes in master database– DB-object metadata converted in Unicode

– Sort order compatibility

• Scripting in Unicode

• Code page override

.

USUS

FrenchFrench

KoreanKorean

8.08.0７７ ..００


Defining your data store

• UCS-2 Server storage

• Data type– Unicode

• NCHAR, NVARCHAR, NTEXT,

• META data – SYSNAME

• N’Unicode’

– DBCS/SBCS• Char, VARCHAR, text


Collation at the database level

•Every database has a collation (default is the server collation)

Create database db_test collate Latin1_General_CI_AI

•Collation can be changed using ALTER DATABASE

SELECT CONVERT(char, DATABASEPROPERTYEX(‘db_test', 'collation'))


Collation at the column level

• Overrides database level collation

CREATE TABLE jobs( job_id smallint IDENTITY(1,1) PRIMARY KEY CLUSTERED, job_desc varchar(50) COLLATE Arabic_CI_AI_KS NOT NULL DEFAULT 'New Position - title not formalized yet',)


Collations by Expressions

SELECT

*

FROM

Table1

WHERE

Field1 = Field2 COLLATE Turkish_ci_ai


The Rules of Precedence for Collations

Explicit B Implicit B Default No Collation

Explicit A Runtime Error Explicit A Explicit A Explicit A

Implicit A Explicit B No Collation Implicit A No Collation

Default Explicit B Implicit B Default No Collation

No Collation Explicit B No Collation No Collation No Collation


Unicode Data FlowServer

OLE/DBOLE/DB ODBCODBC

• 8.0 Client

– ODBC 3.7+/OLEDB

– TDS 8.0, 8.0 Netlibs support Unicode

– Character data converted to/from Unicode on client (Server code page required on client)

– T-SQL batches received in Unicode, parsed in Unicode on Server.

NetlibNetlib

ODSODS

SQL ServerSQL ServerApplicationApplication

NetlibNetlib

• Downlevel client

– DBLIB, ODBC <3.7 clients

– TDS 4.2

– T-SQL batch received in DBCS/ANSI code page of the client

– Translated to Unicode using the Server code page on the Server

DB-LibDB-Lib ODBCODBC

ApplicationApplication

NetlibNetlib

TDS 8.0TDS 8.0 TDS 4.2TDS 4.2


Data Flow

ODBC client to server

• Language event is always in Unicode• Client ACP to Unicode to server

• ‘A’ - Unicode to server character set

• N‘A’ as Unicode

SQL_C_CHARSQL_C_CHAR CHARCHARUnicode Svr CHAR

Client side conversion

char

SQL_C_CHARSQL_C_CHARNCHARNCHAR

SYSNAMESYSNAMEUnicodeUnicode

SQL_C_CHARSQL_C_CHAR CHARCHARbytesNO TRANSLATION

ACP

ACP

SQL_C_WCHARSQL_C_WCHAR CHARCHARUnicode Svr CHAR

char

ClientClient ServerServer


Data Flow

• OLEDB to server: – SSPROP_INIT_AUTOTRANSLATE as

VARIANT_TRUE

• Server code page on the client

DBTYPE_STRDBTYPE_STR CHARCHARUnicode Svr CHAR

Client side conversion

char

DBTYPE_STRDBTYPE_STRNCHARNCHAR

SYSNAMESYSNAMEUnicodeUnicode

DBTYPE_STRDBTYPE_STR CHARCHARbytesVARIANT_FALSE

ACP

ACP

CHARCHARUnicode Svr CHAR

char

ClientClient ServerServer

DBTYPE_WSTRDBTYPE_WSTR


.NET Providers


Data Access

• International T-SQL– NCHAR and N’’– No name strings in date/time– ODBC timestamp– CONVERT with specific style


Data flow•

Data Transformation…..

OLE DB ODBCFixed fieldASCII delimited

OLE DB ODBCFixed fieldASCII delimitedHTML pageRepl. publication

DTS Data Pump

Xforms

Source Destination

In Out

Steps


Client Flow

• Session language syslanguages– Precedence

• Set by ‘set language’

• Set by connection attribute

• Set by user record in syslogins

– Cultural behavior• Language of error messages

• Date format, month name

• Day of week and abbreviations


BCP

– bcp -w : Performs bulk copy operation using Unicode characters.

– bcp -N : Performs the bulk copy operation using the native (database) data types of the data for non-character data, and Unicode characters for character data.


XML

• You can specify an output encoding in a URL.

• XML templates can specify an encoding.

• Unicode by default


Full text

• Allows for word or phrase-based indexing of character data.

• Full-text indexing enables the creation and population of the full-text catalogs, which are maintained outside of SQL Server and managed by the Microsoft Search service.

• Full-text search uses the new Transact-SQL predicates (CONTAINS, CONTAINSTABLE, FREETEXT, and FREETEXTTABLE) to query these populated full-text catalogs.

• With a full-text query, you can perform

– A linguistic search of character data in tables enabled for full-text search.

– A linguistic search operates on words and phrases unlike the LIKE predicate which is used to search character patterns.

• Manipulate to get what you want


ToolsManageability • Unicode based

– SQL-DMO


ToolsWeb Assistant


Backup and Restore

• Restore uses the collation of the source databases

• Verify the collation is support on the instance of SQL Server


Replication PublisherPublisher

DistributorDistributor

Updating Updating SubscriberSubscriber

(immediate updates)(immediate updates)

2PC, RPC2PC, RPC

SubscriberSubscriber SubscriberSubscriber


Analysis Services

OLAP ServerOLAP Server

OLEDB, ADO, XML/AOLEDB, ADO, XML/A

Client TierClient Tier• MD ActiveX ControlsMD ActiveX Controls• MD Extension to OLE DBMD Extension to OLE DB• Office 2000 InterfacesOffice 2000 Interfaces• 3rd Party Clients3rd Party Clients

OLAP Server TierOLAP Server Tier Multidimensional data modeling andMultidimensional data modeling and calculation enginecalculation engine Persistent multidimensional cachePersistent multidimensional cache

OLEDB / ODBCOLEDB / ODBC


Unicode Data Flow in Fringe Areas

• Script usage

– Command line tools

• ISQL utility does not support Unicode input files.

• OSQL -u (Specifies that output_file is stored in Unicode format).

– Query analyzer, save as Unicode / ANSI / OEM.


Resources

• International Features in Microsoft SQL Server 2000 Http://msdn.microsoft.com/library/default.asp

• Arabic Language Support in Microsoft SQL Server 2000 Http://msdn.microsoft.com/library/default.asp

• SQL Server Books On Line

http://msdn.microsoft.com/library/default.asp




22nd International Unicode Conference San Jose, CA September 20021 Going Global With SQL Server 2000 Beverly Sherry Program Manager Global Release Services.

Documents