The development community gains access
to the new features of SQL Server 2012 through the SQL Server Native
Client (SNAC). If the new features are not needed and managed code is a
requirement for data access, ADO.NET can suffice.
Note
ADO.NET is an umbrella label applied to
the .NET functionality that supports connections to a variety of data
sources. Classes within this library supply the programmatic capability
to create, maintain, dispose of, and execute actions against a database.
SQL Server Native Client OLE DB Provider Being Deprecated
The SQL Server Native Client OLE DB
Provider is being deprecated in SQL Server 2012 and will no longer be
available in future releases. Please note that only the SQL Server
Native Client OLEDB Provider is being deprecated at this time and other
OLE DB providers will still be supported until an official announcement
is made on their respective deprecations.
The reasoning behind this move is to
align with the industry for using ODBC as the standard for native
relational data access. This shift not only allows for universal
cross-application compatibility, but this allows developers to
concentrate on one set of APIs for all their native client applications.
This means that if you're developing for standalone SQL Server and
Windows Azure SQL Database, or porting applications to Windows Azure for
the first time, it's all built on the same API so the development
transition is seamless.
Native Client Support for LocalDB
SQL Server 2012 introduced Microsoft SQL
Server Express LocalDB, a lightweight execution mode of SQL Server
Express that allows developers to connect to a SQL Server Engine to
write and test Transact-SQL code without the hassle of having to manage a
full instance of SQL Server.
Metadata Discovery
Thanks to updates to metadata discovery
in SQL Server 2012, applications using SQL Server Native Client will now
get back column or parameter metadata from a query execution identical
to or compatible with the metadata format you specified.
When developing applications that use the SQL
Server Native Client while connected to an earlier version of SQL
Server, the metadata discovery functionality corresponds to the version
of the server.
High-Availability/Disaster Recovery Support
When a database connection is
established to a SQL Server 2012 server, the failover partner is
automatically identified in a mirrored scenario. This information is
used by SNAC to transparently connect to the failover partner if the
principal server fails. The failover partner's identity can also be
supplied as an optional parameter in the database connection string
using the Failover_Partner keyword.
Additionally, you can also use SNAC to take
advantage of the new AlwaysOn Availability Groups feature built on the
mirroring technology. Applications using SNAC can connect to the primary
replica of a given availability group either via a network name or
connection strings much like you do with mirroring.
Applications connecting to an availability group
using SQL Server Native Client can connect using an availability group
listener, which is a virtual network name for the availability group.
This concept should be familiar if you've ever set up a cluster and had
to configure a virtual name for the cluster.
Note
If the connection to the principal server
fails for a transaction, and if a transaction is involved, it will be
rolled back. The connection must be closed and reopened to access the
failover partner, and then any data work must be reapplied. The failover
connection is automatic as long as the failover partner's identity is a
part of the connection object supplied from either the connection
string or a prior successful connection to the server before it went
down.
Improved Date/Time Support
SQL Server 2008 introduced the distinct
date and time data types. These distinct data types provide much more
flexibility for application developers when dealing with date and time
types.
The distinct time type offers precision accurate up to 100 nanoseconds. This precision can be accessed via new types in SNAC: DBTYPE_DBTIME2 for OLE DB providers and SQL_SS_TIME2
for ODBC providers. If an application is written to use time with no
fractional seconds, you can use time(0) columns. Precision of this
magnitude can be extremely useful in process control and manufacturing
applications that require that level of precision.
Other time-related support available is the
ability to use date-time values with timezone information. This feature
is supported through the use of DBTYPE_DBTIMESTAMPOFFSET for OLE DB and SS_TIMESTAMPOFFSET for ODBC types.
Accessing Diagnostic Information in the Extended Events Log
Another new feature in SNAC 11 is that
data access tracing has been updated to make it easier to get diagnostic
information about connection failures from the connectivity ring buffer
and application performance information from the extended events log.
In SQL Server 2012, connection operations via the
SQL Server Native Client will generate a client connection ID. If
connectivity fails, you can access the connectivity ring buffer and find
the ClientConnectionID field to
get diagnostic information about the session and the connection failure.
Note
Data access tracing is intended only for
troubleshooting and diagnostic purposes and may not be suitable for
auditing or security purposes.
ODBC Features
With SQL Server 2012, Microsoft has made
a radical shift on its stance with regard to ODBC. As of this release,
they have decided to push ODBC as the primary method of connectivity and
will be deprecating OLEDB in future releases, although it's still
available in SQL 2012. In this release, the SQL Server Native Client OLE
DB provider is the first of the OLE DB providers that will be
deprecated and will not be available in the next release of SQL Server;
however, it will be supported for the life cycle of SQL Server 2012.
As for the ODBC provider itself, three new
features were added to the ODBC support, which were also added to the
standard ODBC in the Windows 7 SDK:
- Asynchronous execution on connection-related operations
- C Data Type Extensibility
- Calling SQLGetData with a small buffer multiple times to retrieve a large parameter value
Multiple Active Result Sets (MARS)
SQL Server 2012 provides support for
having multiple active SQL statements on the same connection. This
capability includes being able to interleave reading from multiple
result sets and being able to execute additional commands while a result
set is open.
Microsoft guidelines for applications using MARS include the following:
- Result sets should be short-lived per SQL statement.
- If a result set is long-lived or large, then server cursors should be used.
- Always read to the end of the results and use API/property calls to change connection properties.
Note
By default, MARS functionality is not
enabled. Turn it on by using a connection string value — MarsConn for
the OLE DB provider and Mars_Connection for the ODBC provider.
XML Data Types
Much like the current VarChar data type
that persists variable character values, a new XML data type persists
XML documents and fragments. This type is available for variable
declarations within stored procedures, parameter declarations, and
return types and conversions.
User-Defined Types
These types are defined using .NET
Common Language Runtime (CLR) code. This would include the popular C#
and VB.NET languages. The data itself is exposed as fields and
properties, with the behavior being exposed through the class methods.
Large Value Types
Three new data types have been
introduced to handle values up to 2 ˆ 31-1 bytes long. This includes
variables, thus allowing for text values in excess of the old 8K limit.
The data types and their corresponding old types are listed in Table 1.
Table 1 SQL Server 2012 Large Values Types
| varchar(max) |
Text |
| nvarchar(max) |
Ntext |
| varbinary(max) |
Image |
FILESTREAM Support
FILESTREAM is a feature that was
introduced in SQL Server 2008 that allows you to store and access large
binary values, either through SQL Server or by direct access to the
Windows file system. Large binary values are considered any values
larger than 2 gigabytes (GB). This feature allows SQL Server-based
applications to store and manipulate unstructured data, such as images
and documents, on the file system.
Handling Expired Passwords
This feature of SQL Server 2012 enables
users to change their expired password at the client without the
intervention of an administrator.
A user's password may be changed in any of the following ways:
- Programmatically changing the password such that both the old and new passwords are provided in the connection string
- A prompt via the user interface to change the password prior to expiration
- A prompt via the user interface to change the password after expiration
Best Practice
If the old and new passwords are
supplied on the connection string, then ensure that this information has
not been persisted in some external file. Instead, build it dynamically
to mitigate any security concerns.
Snapshot Isolation
The snapshot isolation feature enhances concurrency and improves performance by avoiding reader-writer blocking.
Snapshot isolation relies on the row versioning
feature. A transaction begins when the BeginTransaction call is made but
is not assigned a sequence transaction number until the first T-SQL
statement is executed. The temporary logical copies used to support row
versioning are stored in tempdb.
Note
If tempdb does not have enough space for
the version store, then various features and operations, such as
triggers, MARS, indexing, client executed T-SQL, and row versioning will
fail, so ensure that tempdb has more than enough space for anticipated
uses.
