Home > Articles > Data > SQL Server

In-Process Data Access in SQL Server 2005

  • Print
  • + Share This
Whether you access data from the client, middle tier, or server, when you’re using SQL Server and the .NET Framework, you use the SqlClient data provider. Your data access code is similar regardless of its location, but the .NET Framework 2.0 version of SqlClient contains code to encapsulate differences when you’re programming inside SQL Server and optimize the in-server programming environment. This chapter introduces you to the idea of in-process data access.
This chapter is from the book

Whether you access data from the client, middle tier, or server, when you’re using SQL Server and the .NET Framework, you use the SqlClient data provider. Your data access code is similar regardless of its location, but the .NET Framework 2.0 version of SqlClient contains code to encapsulate differences when you’re programming inside SQL Server and optimize the in-server programming environment.

Programming with SqlClient

Accessing data from outside SQL Server entails connecting to SQL Server through a network library and a client library. When you use the .NET Framework with SQL Server, you use System.Data.dll as the client library and the ADO.NET programming model. ADO.NET is a provider-based model, similar in concept to ODBC, OLE DB, and JDBC. The model uses a common API (or a set of classes) to encapsulate data access; each database product has its own provider. ADO.NET providers are known as data providers, and the data provider for SQL Server is SqlClient. The latest release of SqlClient, installed with .NET Framework 2.0, includes new client-side functionality to take advantage of new features in SQL Server 2005. In addition, SqlClient contains extensions to allow ADO.NET code to be used inside the database itself. Though T-SQL is usually preferred when a stored procedure, user-defined function, or trigger accesses database data, you can also use ADO.NET when writing procedural code in the .NET Framework. The programming model when using SqlClient in .NET Framework stored procedures is similar to client-side code but in-database access is optimized because no network libraries are needed. Let’s start by writing some simple client database code and then convert it to run on the server.

Simple data access code is very similar regardless of the programming model used. To summarize, using ADO.NET and SqlClient as an example:

  1. Connect to the database by instantiating a SqlConnection class and calling its Open method.
  2. Create an instance of a SqlCommand class. This instance contains a SQL statement or procedure name as its CommandText property. The SqlCommand is associated with the SqlConnection.
  3. Execute the SqlCommand, and return either a set of columns and rows called SqlDataReader or possibly only a count of rows affected by the statement.
  4. Use the SqlDataReader to read the results, and close it when finished.
  5. Dispose of the SqlCommand and SqlConnection to free the associated memory, network, and server resources.

The ADO.NET code to accomplish inserting a row into a SQL Server table would look like Listing 4-1.

Listing 4-1: Inserting a row using SqlClient from the client

// Code to insert data from client
//  See chapter 14 for an implementation of
//  the GetConnectionStringFromConfigFile method.
string connStr = GetConnectionStringFromConfigFile();
SqlConnection conn = new SqlConnection(connStr);
conn.Open();
SqlCommand cmd = conn.CreateCommand();
cmd.CommandText = "insert into test values (’testdata’)";
int rows_affected = cmd.ExecuteNonQuery();
cmd.Dispose();
conn.Dispose();

The previous ADO.NET code ignored the fact that an exception might cause the execution of cmd.Dispose or conn.Dispose to be skipped. The preferred and simple way to prevent this from happening is to use the using syntax in C#. One or more object instance declarations are followed by a block of code. The Dispose method is called automatically at the end of the code block. We’ll be using the using construct a lot in the code in this book. Rewritten using this syntax, the code above would look like Listing 4-2.

Listing 4-2: Inserting a row using SqlClient from the client, C# using construct

//code to insert data from client
string connStr = GetConnectionStringFromConfigFile();
using (SqlConnection conn = new SqlConnection(connStr))
using (SqlCommand cmd =
 new SqlCommand("insert into test values (’testdata’)", conn))
{
 conn.Open();
 int rows_affected = cmd.ExecuteNonQuery();
} // Dispose called on cmd and conn here

Other classes in a typical ADO.NET data provider include a transaction class (SqlTransaction) to tie Connections and Commands to a database transaction, a parameter collection (SqlParameterCollection) of parameters (SqlParameter) to use with parameterized SQL queries or stored procedures, and specialized Exception and Error classes (SqlException, SqlErrorCollection, SqlError) to represent processing errors. SqlClient includes all the typical classes; Figure 4-1 shows the object model.

Figure 4-1

Figure 4-1 The SqlClient provider object model (major classes only)

The same basic model is used inside the server to access data in .NET Framework stored procedures. It’s familiar to ADO.NET programmers, and using it inside the server makes it easy for programmers to use their existing skills to write procedures. The big difference is that when you’re writing a .NET Framework procedure, you’re already inside the database. No explicit connection is needed. Although there is no network connection to the database, there is a SqlConnection instance. The difference is in the connection string. Outside the database, the connection string should be read from a configuration file and contains items like the SQL Server instance to connect to (server keyword), the SQL Server login (either User ID and Password keywords or Integrated Security=true), and the initial database (database keyword). The connection string that indicates to SqlClient that we’re already inside the database and the provider should just use the existing database context contains only the keyword "context connection=true". When you specify "context connection=true", no other connection string keyword can be used. Listing 4-3 is the same code as above but executing inside a .NET Framework stored procedure.

Listing 4-3: Inserting a row using SqlClient in a SQLCLR stored procedure

//code to insert data in a stored procedure
public static void InsertRowOfTestData()
{
 string connStr = "context connection=true";
 using (SqlConnection conn = new SqlConnection(connStr))
 using (SqlCommand cmd =
  new SqlCommand("insert into test values (’testdata’)", conn))
 {
  conn.Open();
  int rows_affected = cmd.ExecuteNonQuery();
 }
} 

Note that this code is provided as a stored procedure only to explain how to access data on the server. Not only is the code faster as a Transact-SQL (T-SQL) stored procedure but also SQL Server will check the SQL statements for syntactic correctness at CREATE PROCEDURE time. This is not the case with the .NET Framework stored procedure above. When you execute SQL statements by using SqlCommand, it’s the equivalent of using sp_executesql (a system-supplied store procedure for dynamic string execution of commands) inside a T-SQL stored procedure. There is the same potential for SQL injection as with sp_executesql, so don’t execute commands whose CommandText property is calculated by using input parameters passed in by the procedure user.

This code is so similar to the previous client-side code that if we knew whether the code was executing in a stored procedure on the server or on the client, we could use the same code, changing only the connection string. But a few constructs exist only if you are writing server-side code. Enter the SqlContext class.

  • + Share This
  • 🔖 Save To Your Account