Skip to content

Latest commit

 

History

History
76 lines (42 loc) · 5.75 KB

readme.adoc

File metadata and controls

76 lines (42 loc) · 5.75 KB
Raw

Data Serialization

Introduction

Concepts you will gain experience with:

  1. Configuring your cache to perform PDX Auto-Serialization

  2. Working with multiple version of domain objects in PDX

  3. Learn to use the PdxInstance API

Estimated completion time: 40 minutes

Configuring the Cache for PDX Auto-Serialization

In this section, you will focus on configuring PDX Auto-Serialization in both the client and server configurations.

  1. (TODO-01) To begin, open the Customer class definition in the io.pivotal.bookshop.domain package. Take a moment to examine how it’s written. Note first of all that it does not implement java.io.Serializable at all. Also note that one of the fields is an Address instance that represents various attributes related to a customer’s address (ex. city, state, zip code, etc). If you like, you can also open this Address class (in the same package) and note that it also does not implement java.io.Serializable. In this condition, the only way to ensure these objects are serialized is to use one of the PDX Serialization techniques. Note also that both the Customer and Address classes have default constructors, which are a requirement for PDX Serialization.

  2. (TODO-02) Next, open the cluster/cluster.xml file. Your task is to add the appropriate configuration to enable PDX Serialization in the server cache.

  3. (TODO-03) Add an attribute to the pdx element signaling that the server should not de-serialize objects.

  4. cd into the cluster folder, launch gfsh, then use gfsh commands to start a locator and two servers.

    Note
    		 Do not use the --classpath argument when starting servers. It is important in observing the benefit of PDX Serialization that the domain classes not be on the classpath of the servers.

  5. (TODO-04) Open the src/main/resources/clientCache.xml file and make a similar modification to the definition to support PDX Auto-Serialization. Do not set the attribute to force client to read as a serialized object. We actually do want the PDX de-serialization to take place on the client.

  6. (TODO-05) On the client side, we must also specify a serializer class name. Add a class-name xml element and specify the fully qualified class name for the ReflectionBasedAutoSerializer. In addition, configure a parameter to the auto-serializer called classes that registers the classes that should be auto-serialized. You can use wild cards to be sure you get both the Customer class and Address class in the io.pivotal.bookshop.domain package.

  7. (TODO-06) Locate the CustomerLoader class in the io.pivotal.bookshop.buslogic package and run it. This will load 3 customers into the Customer region on the server.

  8. Return to gfsh and execute a query to select values from the Customer region. The output should resemble the following:

custNum | firstName | lastName |                     primaryAddress                  | myBookOrders
------- | --------- | -------- | --------------------------------------------------- | -------------------------
5598    | Kari      | Powell   | class org.apache.geode.pdx.internal.PdxInstanceImpl | class java.util.ArrayList
6024    | Trenton   | Garcia   | class org.apache.geode.pdx.internal.PdxInstanceImpl | class java.util.ArrayList
5543    | Lula      | Wax      | class org.apache.geode.pdx.internal.PdxInstanceImpl | class java.util.ArrayList

Working with PDX Domain Object Versions

In this section, you’ll begin to understand the power of PDX as you modify the definition of the Customer class and see that multiple versions of the class definition can be used within the GemFire distributed system at the same time.

  1. (TODO-07) To begin, open the Customer class again and add a telephoneNumber field of type String. Also, add this fieldname to the @ToString annotation.

  2. (TODO-08) Open the NewCustomerClient class in the io.pivotal.bookshop.buslogic package. In the main() method. Write the code to create a new Customer instance. Be sure to set the new telephoneNumber property. Save the entry with the key 9999.

  3. (TODO-09) Add code to retrieve the newly inserted Customer entry (key: 9999) and print it out (log it).

  4. Run NewCustomerClient to insert this new record into GemFire.

  5. Return to gfsh and re-issue the query command. This time, note that there is the additional entry for key 9999. Notice also that there is a new field displayed for telephoneNumber. Note that the first three entries now show this field value as null, which is the expected behavior.

Using PdxInstance

In this last section, you will gain familiarity working with the PdxInstance class. This class offers the ability to fetch and manipulate objects without necessarily having their domain definitions on the classpath. With PdxInstance, one can also de-serialize only the fields that are required to perform necessary processing.

  1. (TODO-10) Open the PdxInstanceClient class in the io.pivotal.bookshop.buslogic package. Add the necessary code to fetch the entry for key 9999, processing it as a PdxInstance. Extract and print just the telephoneNumber field.

    Tip
    		 It may be a good idea to add a test to ensure the instance you got back is an instance of PdxInstance and print out a notice if it isn’t.

  2. (TODO-11) Return to the clientCache.xml file and modify the attribute of the pdx element to tell the client cache not to de-serialize entries into Customer objects. This is necessary if you intend to process your entries as PdxInstance objects.

  3. Run the PdxInstanceClient and verify that you were able to obtain the telephoneNumber field and that it is the correct value. It should be the value you set in the prior section when you created the new Customer entry.

Congratulations!! You have completed this lab.