Best Practices developing applications: Part 1

There are many reasons why we follow best practices. My own thoughts have changed over the course of my career. Early in my career as a junior developer I strictly followed the patterns and practices of my seniors and leads, and I assumed that their advice would be better than my own.

But with the passage of time my perceptions started to change because of research. As we have to research when our application consumes lot of time in executing and also when its scalability becomes a question. Before doing so I always used DataSet to get the data from the database. Whether it is for large application or too small, but then I found something was doing wrong.

DataSet is a collection of in memory tables and datareader provides the ability to expose the data from database.

Both are very widely used in asp.net applications to get/fetch the data from the database. But for a scalable, fast, and reliable application one has to know the best practices in developing application.

DataSet

We should use when the application is:

Windows application
Not too large data
Returning multiple tables
If, to be serialized
Disconnected architecture
To return from a WCF service
To send across layers
Caching the data
Do not need to open or close connection

Code snippet using dataset:

public DataSet GetRecord(Guid id, string procedureName)
{
    DataSet resultSet = new DataSet();
    SqlConnection connection = new SqlConnection(System.Configuration.
    ConfigurationManager.ConnectionStrings["ConnectionString"].ConnectionString);
    SqlCommand command = new SqlCommand(procedureName, connection);
    command.CommandType = CommandType.StoredProcedure;
    command.Parameters["pID"].Value = id.ToString();
    IDataAdapter adapter = new SqlDataAdapter(command);
    try
    {
        adapter.Fill(resultSet);
    }
    catch (Exception ex)
    {
        throw new PerformanceException(ex.Message, ex.InnerException);
    }
    return resultSet;
}

DataReader

DataReader is a stream which is readonly and forward only. It fetches the record from databse and stores in the network buffer and gives whenever requests. DataReader releasese the records as query executes and do not wait for the entire query to execute. Therefore it is very fast as compare to the dataset. It releases only when read method is called.

Its usages:

Web application
Large data
Returning multiple tables
For Fast data access
Needs explicitly closed
Output parameter value will only available after close
returns only a row after read

Code sample for DataReader:

 public SqlDataReader GetRecord(Guid id, string procedureName)
{

    SqlDataReader resultReader = null;
    SqlConnection connection = new SqlConnection(
      ConfigurationManager.ConnectionStrings["ConnectionString"].ConnectionString);
    SqlCommand command = new SqlCommand(procedureName, connection);
    command.CommandType = CommandType.StoredProcedure;
    command.Parameters["pID"].Value = id.ToString();
    try
    {
        connection.Open();
        resultReader = command.ExecuteReader(CommandBehavior.CloseConnection);
    }
    catch (Exception ex)
    {
        if (resultReader != null || connection.State == ConnectionState.Open)
        {
            resultReader.Close();
            connection.Close();
        }
        throw new PerformanceException(ex.Message, ex.InnerException);
    }

    return resultReader;
}

DataSet vs. DataReader