Caching the Data for Better Performance





Caching the Data for Better Performance

Modern servers, even the cheap ones, generally have a ton of memory, much of which goes unused. The data that you're sucking out of the database and serving to hundreds or thousands of users over and over might not change much, but it does take a bit of time to search for it and extract it from the database. That means reading data off of a hard drive and dragging it through the database's process and then piping it through drivers to .NET so you can display it. Why not store it in memory? It's considerably faster to retrieve the data from memory.

The System.Web.Caching.Cache class provides us with a powerful means to keep objects in memory so that they can be quickly retrieved and used throughout our application.

Part of the power of this class is its ability to decide when the object in memory is no longer needed. It makes this decision based on the type of cache dependency that you choose. You can base the dependency on changes to a file or directory, which is great if you're using some file-based resource, but it's a problem for us because our data isn't coming from a file (not in the literal sense, anyway).

We can also kill off our cached items after a certain amount of time has passed or if the object hasn't been accessed for a certain amount of time.

All these methods make data caching difficult because our data might be changed in the meantime by another page or some other process. Obviously we want only the current data to be served to people visiting our site. The key to maintaining this data integrity while caching our data is to only access the database via our data class.

To cache the data, we'll add a few lines of code to our constructor and our Update() and Delete() methods. We don't have to add any code to the Create() method because at that point, we have no idea if the data will be retrieved by some other page or process.

Cached objects are organized and retrieved by a key, much in the same way that you find the right portion of a query string or access a column by name from a DataReader. We'll name our cached customer objects "UberCustomer" plus the primary key of the record we retrieve. So for example, if my customer record's CustomerID column has a value of 216, the cached object will be named "UberCustomer216."

We'll start by adding a single "if" statement to our constructor, checking to see if the cached object exists. If it does, we'll load those values into our properties. If it doesn't exist, we'll get the data from the database and insert it into the cache. The revised constructor looks like the code in Listing 5.9.

Revised constructor with caching

C#

public Customer(int CustomerID)
{
  HttpContext context = HttpContext.Current;
  if ((context.Cache["UberCustomer" + CustomerID.ToString()] == null))
  {
    SqlConnection connection = new SqlConnection(_ConnectionString);
    connection.Open();
    SqlCommand command = new SqlCommand("SELECT CustomerID, "
      + "LastName, FirstName, Address, City, State, Zip, "
      + "Phone, SignUpDate WHERE CustomerID = @CustomerID", connection);
    command.Parameters.AddWithValue("@CustomerID", CustomerID);
    SqlDataReader reader = command.ExecuteReader();
    if (reader.Read())
    {
      _CustomerID = reader.GetInt32(0);
      _LastName = reader.GetString(1);
      _FirstName = reader.GetString(2);
      _Address = reader.GetString(3);
      _City = reader.GetString(4);
      _State = reader.GetString(5);
      _Zip = reader.GetString(6);
      _Phone = reader.GetString(7);
      _SignUpDate = reader.GetDateTime(8);
    }
    else PopulateDefault();
    reader.Close();
    connection.Close();
    context.Cache.Insert("UberCustomer" +
    _CustomerID.ToString(), this, null, 
    DateTime.Now.AddSeconds(60), new TimeSpan.Zero);
  }
  else
  {
    Customer customer = (Customer)context.Cache["UberCustomer" +
      CustomerID.ToString()];
    _CustomerID = customer.CustomerID;
    _LastName = customer.LastName;
    _FirstName = customer.FirstName;
    _Address = customer.Address;
    _City = customer.City;
    _State = customer.State;
    _Zip = customer.Zip;
    _Phone = customer.Phone;
    _SignUpDate = customer.SignUpDate;
  }
}


VB.NET

Public Sub New(CustomerID As Integer)
   Dim context As HttpContext = HttpContext.Current
   If context.Cache(("UberCustomer" + CustomerID.ToString())) Is Nothing Then
      Dim connection As New SqlConnection(_ConnectionString)
      connection.Open()
      Dim command As New SqlCommand("SELECT CustomerID, LastName, "_
        & "FirstName, Address, City, State, Zip, Phone, "_
        & " SignUpDate WHERE CustomerID = @CustomerID", connection)
      command.Parameters.AddWithValue("@CustomerID", CustomerID)
      Dim reader As SqlDataReader = command.ExecuteReader()
      If reader.Read() Then
         _CustomerID = reader.GetInt32(0)
         _LastName = reader.GetString(1)
         _FirstName = reader.GetString(2)
         _Address = reader.GetString(3)
         _City = reader.GetString(4)
         _State = reader.GetString(5)
         _Zip = reader.GetString(6)
         _Phone = reader.GetString(7)
         _SignUpDate = reader.GetDateTime(8)
      Else
         PopulateDefault()
      End If
      reader.Close()
      connection.Close()
      context.Cache.Insert("UberCustomer" + _CustomerID.ToString(), _
        Me, Nothing, DateTime.Now.AddSeconds(60), TimeSpan.Zero)
   Else
      Dim customer As Customer = _
        CType(context.Cache(("UberCustomer" + CustomerID.ToString())), _ Customer)
      _CustomerID = customer.CustomerID
      _LastName = customer.LastName
      _FirstName = customer.FirstName
      _Address = customer.Address
      _City = customer.City
      _State = customer.State
      _Zip = customer.Zip
      _Phone = customer.Phone
      _SignUpDate = customer.SignUpDate
   End If
End Sub

We start the new version of the constructor by checking to see if an existing cache object corresponds to the record we're looking for. Because our class has absolutely no clue that it's being used in a Web application, we first create an HttpContext object to reference, in this case, HttpContext.Current, which provides a reference to the current request.

If there is no cached object, everything proceeds as before, except for the very last line. We call the Insert() method of the cache object, which takes a number of parameters. (There are a number of overloads for the Insert method, but this one offers the most control for our purposes. Consult the .NET documentation for more information.)


context.Cache.Insert("UberCustomer" + _CustomerID.ToString(), this, null, DateTime.Now
.AddSeconds(60), TimeSpan.Zero);

The first parameter is a string to name the cache entry. As we mentioned earlier, it's a combination of the "UberCustomer" and the CustomerID value. The second parameter is this (or Me in VB), which is the instance of the class itself. That means that all of the values assigned to the class' properties are stored in memory. The third parameter is for a CacheDependency object, and in our case we're using null (Nothing in VB) because we're not tying any dependency to the cached object.

The fourth parameter is the time at which we want the cached object to be removed from memory, which is an absolute expiration time. The fifth parameter is a sliding expiration time expressed as the TimeSpan that passes without the cached object being accessed. That means an object could live indefinitely if it's accessed over and over. Because we've already set an absolute expiration, we must set this to a TimeSpan object that indicates zero time.

You must experiment with these values to decide how much memory you want to use (see Chapter 15, "Performance, Scalability, and Metrics"). If you write a number of different data classes similar to this one, you may want to store a value in web.config that indicates the number of seconds (or minutes, hours, or whatever you want) so that you can change the setting all from one place.

If the object has been cached, it's easy enough to retrieve those values and assign them to our private class members. We create a new Customer object and fill it with the cached version. We have to cast the object to the Customer type because the type returned by the Cache object is System.Object.

This is a point of confusion for some developers because we're creating an instance of the class from within the class and then assigning its properties to the private members of the class in which we're working.

Getting this cached data will save many trips to the database, but we need to devise a way to make sure that we always have current data. If another user loads the data and changes it by calling Update() or deletes it with the Delete() method, we must remove the cached data so that it is sought from the database instead of being loaded from memory. This is easy enough with a private method that uses the cache's Remove() method. Listing 5.10 demonstrates the cache removal.

Private DeleteCache() method

C#

private void DeleteCache()
{
    if (HttpContext.Current.Cache["UberCustomer"
       + _CustomerID.ToString()] != null)
       HttpContext.Current.Cache.Remove("UberCustomer"
          + _CustomerID.ToString());
}


VB.NET

Private Sub DeleteCache()
    If Not (HttpContext.Current.Cache(("UberCustomer" & _
       _CustomerID.ToString())) Is Nothing) Then
       HttpContext.Current.Cache.Remove(("UberCustomer" & _
          _CustomerID.ToString()))
    End If
End Sub

Again, we reference the HttpContext.Current object. First we see if the object exists, and if it does, we call the Remove() method, which looks for the object by its key name.

We'll need to call the DeleteCache() method from both the Update() and Delete() methods. It's as simple as adding one line to both of the methods: DeleteCache().

As long as we access the database through this class only, we will always have current data.


     Python   SQL   Java   php   Perl 
     game development   web development   internet   *nix   graphics   hardware 
     telecommunications   C++ 
     Flash   Active Directory   Windows