12c SOA - Oracle BPEL Dehydration

BPEL dehydration process:

• BPEL dehydration process is the action from the BPEL engine to compress and store the BPEL instances into the database. The incoming messages are stored in the database for processing later by worker threads.
• Storing the current status of the BPEL process(i.e. long running process, asynchronous process) into the database tables is known as dehydration.
• SOA_INFRA schema is the dehydration store which contains tables to hold the meta data of the process.

Why Dehydration

• Long running process or processes waiting for response consumes memory and CPU While waiting for the response the bpel engine can store the process, thus freeing up server resources.
• To increase both Reliability and Scalability of the BPEL process.
• You can also use it to support clustering and failover.
• Over the life cycle of the BPEL instance, the instance with the current state of execution may be saved in database.

Note: BPEL Activity is dehydrated immediately after execution and recorded in the dehydration store. It provides better failover protection, but may impact performance if the BPEL process accesses the dehydration store frequently.

Synchronous Process

• Process gets dehydrated only at the end of the process.
• Using Dehydrate activity we can explicitly dehydrate process state if required.

Asynchronous Process

• Automatically dehydrated the process based on the activities used.

When dehydration occurs:

When the BPEL instance encounters a mid-process breakpoint activity (not including the initial receive). The activities are used : 

• a mid-receive activities(without create instance). 
• Pick (onMessage & onAlarm) activity
• just before wait activity (depends on the duration)
• Dehydrate(11g) / Checkpoint(10g)
• Commit(Java embedded)
• a (implicit) transaction is committed

That is where an existing BPEL instance must wait for an event, which can be either a timer expiration or message arrival. When the event occurs (the alarm expires or the message arrives), the instance is loaded from the dehydration store and execution is resumed. This type of dehydration occurs only in durable processes, which have mid-process breakpoint activities. A transient process does not have any mid process breakpoint activities.

When the BPEL instance encounters a non-idempotent activity

When Oracle BPEL Server recovers after a crash, it retries the activities in the process instance. However, it should only retry the idempotent activities. Therefore, when Oracle BPEL Server encounters a non-idempotent activity, it dehydrates it. This enables Oracle BPEL Server to memorize that this activity was performed once and is not performed again when Oracle BPEL Server recovers from a crash.

Idempotent activities are those activities where the result is the same irrespective of no. of times you execute the process.

Repeated invocations have the same effect as one invocation.

E.g. Read only services, Receive activity

When RequiresNew BPEL transaction is completed or When the BPEL instance finishes

At the end of the BPEL process, Oracle BPEL Server saves the process instance to the dehydration store, unless you explicitly configure it not to do so. This happens to both durable and transient processes.

Note: A BPEL invoke activity is by default (true) an idempotent activity, meaning that the BPEL process does not dehydrate instances immediately after invoke activities. Therefore, if idempotent is set to true and Oracle BPEL Server fails right after an invoke activity executes, Oracle BPEL Server performs the invoke again after restarting. This is because no record exists that the invoke activity has executed. This property is applicable to both durable and transient processes. If idempotent is set to false, the invoke activity is dehydrated immediately after execution and recorded in the dehydration store. If Oracle BPEL Server then fails and is restarted, the invoke activity is not repeated, because Oracle BPEL Process Manager sees that the invoke already executed.

Idempotent vs non idempotent activities
Idempotent:Activities which can be retried are called idempotent activities. Examples are : assign activity, invoke activity etc. So if the BPEL process is recovered from the dehydration store , then these activities are retried again (idempotent activities are not stored in dehydration store if there is no dehydration activity like checkpoint() ) .
Non idempotent: Activities are those which causes the BPEL process to dehydration and so there state is saved into dehydration store. If the BPEL process is recovered from the dehydration store, these activities are not retried . Examples are : wait activity, mid point receive activity, pick activity , checkpoint() etc.

So when you specify the idempotent property as "false" in the partner link of a BPEL process , this becomes non idempotent and cause the BPEL process for dehydration and so a commit occurs . This is another way of ending the current transaction of a BPEL process and start a new one.

Types of BPEL Process
When and how dehydration occurs differ based on process types. Based on the dehydration we can categorize process in to 2 categories

Transient Process: Oracle BPEL server dehydrates the process instance only at the end of the process. If server crashes in middle of the running process instance, the instances will not be visible in EM
Durable Process: Oracle BPEL Server dehydrates the process instance in-flight at all midprocess breakpoint and non-idempotent activities, plus the end of the process. When the server crashes, this process instance appears in Oracle BPEL Control up to the last dehydration point (breakpoint activity) once the server restarts. If the server crashes before the process instance reaches the first midprocess breakpoint activity, the instance is not visible in Oracle BPEL Control after the server restarts.

Important Notes:

• Remember that theses dehydration statement must be avoided in synchronous BPEL process.
• Idempotence is a mathematical term that basically means that calling a function multiple times doesn’t change the result for example f(f(x))=f(x).
• What does it mean for messaging service system ? If a service receives the same message again, it should be able to handle it without changing the state of the system. For example; in a bank scenario you wouldn’t want that withdraw message to be processed more than once !
• An idempotent activity in BPEL (for example, an assign activity or an invoke activity) is an activity that can be retried
• To ensure data consistency, the dehydration database is done using the same transaction context in which the BPEL process is executed. However, BPEL engine use a separate transaction context for the audit logging information, which aids (a lot) for debugging failed instance.

Skipping dehydration

• Skipping dehydration will increase performance with a cost of auditing.
• By Default, all BPEL processes are dehydrated regardless of whether they are Synchronous or Asynchronous process.
• For Synchronous processes that do not need to be durable, you can turn off the dehydration mechanism by setting the following properties at process or engine level:

1. Set inMemoryOptimization to true.
2. Set completionPersistPolicy property to faulted or Off.

• Asynchronous process invocation messages are always saved to the dehydration store.
• Durable process are always dehydrated regardless of the settings of the persistence properties.

Dehydration Tables
Cube_Instance: Stores the information about the composite instance that gets created.
Cube_scope: Stores information about the scopes and variables declared and used.
DLV_Message: All the incoming payload details are stored.
Invoke_Message: All the outgoing message payload details are stored.
Audit_Trail: Used to store information of the xml that is rendered in EM console.