The simpleEventSource adapter is configured to record events. The recording happens as events flow out of the adapter. The eventStream channel is configured to play back events. The playback happens where events flow into the channel.

The following graphic shows the EPN of the event record and playback example to demonstrate where you can record events and where you can play events back.

Figure 15-1 Configuring Record and Playback in an EPN

Description of "Figure 15-1 Configuring Record and Playback in an EPN"

Berkeley DB is a fast, scalable, transactional database with industrial grade reliability and availability.

When you record events, by default the Oracle Stream Analytics server stores them in Berkeley DB, which is a persistent event store that is bundled with the Oracle Stream Analytics server. For more information about Berkeley DB, see:

http://www.oracle.com/technetwork/database/database-technologies/berkeleydb/overview/index.html.

When you deploy an application that is configured to use the record and playback feature, the Oracle Stream Analytics server creates the database schema and an instance of Berkeley DB in the following directory.

/Oracle/Middleware/my_oep//user_projects/domains/domainname/servername/bdb

You can use the default Berkeley database configuration as is. You only need to make configuration changes to customize the location of the Berkeley database instance or to tune performance. See Configure Berkeley DB for information about how to configure Berkeley DB.

You can use the event store API to query a store for past events given a record time range and the component from which the events were recorded. The actual query you use depends on the event repository provider; for example, you would use Oracle CQL for the default persistent event store provider included with Oracle Stream Analytics. You can also use these APIs to delete old events from the event store.

You can configure recording for any component in the EPN that produces events, such as processors, adapters, channels, and event beans. Processors and channels always produce events.

Adapters and event beans must implement the EventSource interface.

You can configure events from different components in the EPN to be stored in different persistent stores, or that all events go to the same store. Only events that are output by the component are recorded.

You enable the recording of events for a component by updating its configuration file and adding the record-parameters element. Using the child elements of record-parameters, you specify the event store to which the events are recorded, an initial time period when recording should take place, the list of event types you want to store, and so on.

After you deploy the application and events start flowing through the network, recording begins either automatically because you configured it to start at a certain time or because you dynamically start it using administration tools. For each component you have configured for recording, Oracle Stream Analytics stores the events that flow out of it to the appropriate store along with a time stamp of the time it was recorded.

You can configure playback for any component in the EPN: processors, adapters, streams, and event beans. Typically the playback component is a stage later in the network than the stage that recorded the events.

You enable the playback of events for a component by updating its configuration file and adding the playback-parameters element. Using the child elements of playback-parameters, you specify the event store from which the events are played back, the list of event types you want to play back (by default all are played back), the time range of the recorded events you want to play back, and so on. By default, Oracle Stream Analytics plays back the events in a time accurate manner. However, you can configure that the events get played back either faster or slower than they originally flowed out of the component from which they were recorded.

After you deploy the application and events start flowing through the network, you must start the playback with Oracle Stream Analytics Visualizer or wlevs.Admin. Oracle Stream Analytics reads the events from the persistent store and inserts them into the appropriate place in the EPN.

When a component receives a playback event, the playback event looks exactly like the original event. If a downstream component is configured to record events, then Oracle Stream Analytics records the arriving playback events and real-time events.

You can use the default Berkeley DB configuration as is. You only need to make configuration changes to customize the location of the Berkeley database instance or to set the cache size to tune performance.

1. Stop the Oracle Stream Analytics server instance, if it is running.
2. Open the server config.xml file for editing.

   The file is located in /Oracle/Middleware/my_oep/user_projects/ domains/<domain_name>/<server_name>/config.

3. Edit the bdb-config element in the config.xml file.

   The following example shows a fully configured bdb-config element.

   <bdb-config>
       <db-env-path>bdb</db-env-path>
       <cache-size>1000</cache-size>
   </bdb-config>
   

   Table 15-1 lists the bdb-config child elements.

   Table 15-1 Child Elements of bdb-config

   Child Element	Description
   db-env-path	Specifies the subdirectory in which OracleEvent Processing server creates the Berkeley DB instances relative to the config directory of your server. Default: bdb
   cache-size	Specifies the amount of memory, in bytes, available for Berkeley DB cache entries. You can adjust the cache size to tune Berkeley database performance. The cache size must be a power of 2, but it is otherwise limited only by available memory and performance considerations. Default: je.maxMemoryPercent * JVM maximum memory.

4. Restart the Oracle Stream Analytics server.

You can configure any processor, adapter, channel, or event bean in your application to record events.

1. Open the component configuration file and add a record-parameters child element to the component you want to configure to record events, as follows:

       <csv-adapter>
         <name>StockTradeCSVAdapter</name>
         <record-parameters>
             ...
         </record-parameters>
       </csv-dapter>
   

2. Add child elements to record-parameters to specify a data set name, the list of one or more events to be stored, the recording start and stop times, and so on:

   <csv-adapter>
     <name>StockTradeCSVAdapter/name>
     <record-parameters>
       <dataset-name>recplay_sample</dataset-name>
       <event-type-list>
         <event-type>TradeEvent</event-type>
       </event-type-list>

       <time-range>
         <start>2010-01-20T05:00:00</start>
         <end>2010-01-20T18:00:00</end>
       </time-range>

     </record-parameters>
   </csv-adapter>
   

yyyy-mm-ddThh:mm:ss

  <time-range>
    <start>2010-01-20T05:00:00</start>
    <end>2010-01-20T18:00:00</end>
  </time-range>

yyyy-mm-ddThh:mm:ss

hh:mm:ss

  <time-range-offset>
    <start>2010-01-20T05:00:00</start>
    <duration>03:00:00</duration>
  </time-range-offset>

You can configure any processor, adapter, channel, or event bean in your application to play back events. The component must downstream from the recording component so that the playback component can receive the events and play them back.

1. Open the component configuration XML file and add a playback-parameters child element to the component you want to configure to playback events. For example, to configure a channel called eventStream:

   <channel>
     <name>eventStream</name>
     <playback-parameters>
      ...
     </playback-parameters>
   </channel>
   

2. Add child elements to playback-parameters to specify a data set, one or more events to be played back, and so on. For example:

       <channel>
         <name>eventStream</name>
         <playback-parameters>
             <dataset-name>recplay_sample</dataset-name>
             <event-type-list>
                 <event-type>SimpleEvent</event-type>
             </event-type-list>
         </playback-parameters>
       </channel>
   

yyyy-mm-ddThh:mm:ss

  <time-range>
    <start>2010-01-20T05:00:00</start>
    <end>2010-01-20T18:00:00</end>
  </time-range>

yyyy-mm-ddThh:mm:ss

hh:mm:ss

  <time-range-offset>
    <start>2010-01-20T05:00:00</start>
    <duration>03:00:00</duration>
  </time-range-offset>

After you configure the record and playback functionality for the components of an application, and you deploy the application to Oracle Stream Analytics, the server starts to record events only when you have configured explicit star and stop information in the configuration file.

For example, if you included the following element in a component configuration, then recording starts on January 20, 2010 at 5:00 am:

<time-range>
    <start>2010-01-20T05:00:00</start>
    <end>2010-01-20T18:00:00</end>
</time-range>

To enable the recording and playback of events, use Oracle Stream Analytics Visualizer or wlevs.Admin. Once recording and playback are enabled, they start and stop according to their configuration settings.

For more information, see:

• Start Playback in Administering Oracle Stream Analytics

• Stop Playback in Administering Oracle Stream Analytics

Visualizer and wlevs.Admin use managed beans (MBeans) to dynamically start and stop event recording and playback and manage the event store configuration. A managed bean is a Java bean that provides a Java Management Extensions (JMX) interface. JMX is the Java EE solution for monitoring and managing resources on a network. You can create your own administration tool and use JMX to manage event store functionality with com.bea.wlevs.management.configuration.StageMBean.

For more information, see:

• Java API Reference for Oracle Stream Analytics.