Developing a Message Driven Architecture with Spring

Developing a Message Driven Architecture
with Spring – an Overview
Silicon India Java Conference, Hyderabad
October 15, 2011
•
Technology Specialist with Deloitte Consulting India, practicing Software
Engineering for more than a decade now
•
Defined Technical architecture for multiple strategic projects for clients across
domains, but predominantly in Financial Services Industry
•
Deep experience in creating application development frameworks and solution
sets using Java
•
Leads the competency development of Java Community of Practice at Deloitte
Consulting India
•
Trainer and a speaker - conducted multiple tech-talks predominantly around
Architecture, Design and Software Engineering best practices.
•
Member of Scrum Alliance and a certified Scrum Master
-2-
Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
Introduction – Karthik Banda
Agenda
•
Architectural Context for Integration
•
Architectural Patterns in Discussion
• Event Driven Architecture (EDA)
• Spring as EDA enabler
• Message Driven Architecture
Java Message Service (JMS)
Message Exchange Patterns
Message Driven Beans in EJB 3 and Message Driven POJO’s
Realizing Message driven designs with Spring
•
Spring Task Execution and Scheduling
•
Spring Integration
•
Q&A
-3-
Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
•
•
•
•
Common Architectural contexts and respective Choices
Integration and communication styles–
• Online / Real Time Integration – SOA Enablers – ESB / MOM / JMS / XML/RPC/JCA
and Web Services
• Offline Integration – Scheduled Batch processes, Polling
• Synchronous Communication – RMI / RPC/JCA /Web Services
• Asynchronous Communication – Service Activation / JMS / Asynchronous webservice calls
• Message Channeling, Routing, Aggregation – ESB or could be Spring Integration
-4-
Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
•
Event Driven Architecture (EDA) – in a Nutshell
•
One of the key design principle is - Low Coupling and High Cohesion
•
EDA gives the flexibility to define clearly the boundaries and responsibilities of
classes in collaboration but keeping their level of dependency to the minimum
•
Example : “What Should I do on successful booking of a ticket”
•
EDA is all about transmitting events between different components in a system
and clearly defining a way to handle them
•
An “Event” generally represents a change in the state of a system
•
EDA avoids any coupling between subsystems as it takes away the normal
command and control scheme i.e. one class calling a method of another
“known” class and waiting for it to respond.
-5-
Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
•
Example : “ Booked a Ticket Successfully”
•
Events are broadcasted whenever they occur and they are fine grained
•
Events typically are handled Asynchronously and the Caller is not aware of
Callee
•
Fundamental principle of “Services” to be stateless can be easily achieved
•
No blocking times – Events no longer takes place one at a time, and the order
of events is no longer important.
-6-
Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
EDA – Characteristics
EDA – Spring - Sending Application Events
1. Publisher injected by the container – Setter Injection in this case
-7-
Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
2. Use the publisher to send events at runtime.
EDA – Spring - Handling Application Events
1. Implement the ApplicationListener interface with a parameterized type.
-8-
Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
2. Define the bean, and it will be invoked at runtime when that type of Event
occurs.
•
One of the common architectural choice to integrate disparate systems in a
distributed environment is Messaging
•
Uses a Middleware to facilitate messaging i.e. to queue sent messages and
handle them – Message Oriented Middleware (MOM)
•
Message Producers and Consumers are not aware of each other
•
JMS is a MOM standard and is a vendor neutral specification for messaging
•
Features of JMS include
•
Point to Point Messaging (Queues) (1sender-1receiver)
•
Publish and Subscribe (Topics) (1publisher – N subscribers)
•
Reliable Message Delivery
•
Message Persistence
•
Transactional
•
Structured Messaging
-9-
Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
Message Driven Architecture (MeDA)- JMS - in a Nutshell
•
Point to Point Messaging Domain – Messages are sent to queues and are
delivered only once and that too to a single consumer. Queues retain
messages until consumed or expired
•
Publish and Subscribe Messaging Domain – Messages are sent to topics
and are delivered to all the subscribers. Ideal for publishing business events
and it is one flavor of implementing EDA.
- 10 -
Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
MeDA – Messaging Domains / Models
MeDA – Implementation choice – Message Driven Beans
Characteristics of Message Driven Beans
•
Message Driven beans are state less server side Java objects for handling
JMS Messages and processing them asynchronously
•
Developed as part of EJB 2.0 specification
•
Highly scalable server side components as they are easily managed and/or
distributed
•
Mandates the need for an EJB container / JEE application Server
•
Heavy life cycle as part of EJB specification (Nothing changed in EJB 3 )
•
Each MDB listens to a single destination
•
Cannot be part of domain model as they are otherwise are not reusable
•
Normal leverage is only to the extent of “Services Activation”
- 11 -
Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
Disadvantages of Message Driven Beans
MeDA – Implementation choice – Message Driven POJO
Characteristics of Message Driven POJO
•
These can be part of domain model as they are normal java classes with no
additional baggage
•
These classes are JMS unaware
•
Developer need not worry about JMS
•
Easily testable
•
Developers think in terms of Command – Control instead of Messaging and
Asynchronous communication as these POJO’s are not treated separately as
“Special Objects”
•
Needs plumbing code to leverage JMS resources. Needs more attention to
“Separation of Concerns”.
•
No standard interfaces are implemented by limiting their power of interface
injection
- 12 -
Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
Disadvantages of Message Driven POJO’s
MeDA – Spring – Sending JMS Messages
1. Abstract JMS Administered objects into a JmsTemplate
2. Inject an instance of Spring's JmsTemplate into the sending class
- 13 -
Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
3. Provide the JMS ConnectionFactory in the JmsTemplate bean definition
MeDA – Spring – Listening for JMS Messages
1. Define a “listener-container” within the context and specify the destination that
needs to be looked at for receiving messages
- 14 -
Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
2. Point to any POJO and specify a listening method to implicitly create a
MessageListenerAdapter
•
AMQP is an open standard application layer protocol for message-oriented
middleware
•
One level ahead of JMS – unlike JMS, AMQP is a wired protocol
•
Dictates the format of messages during data transfer across networks
- 15 -
Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
MeDA – Spring – Advanced Message Queuing Protocol
MeDA – Spring – Sending/Receiving AMQP Messages
1. Use AmqpTemplate which accepts Exchange and Routing Key
- 16 -
Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
2. Point to any POJO and specify a listening method to implicitly create a
MessageListenerAdapter for AMQP messages.
•
Better to use events/messaging and only poll when you have to
•
Lifecycle interface – start(), stop() and isRunning() - This supports any basic
back ground task and a common usage is polling
•
SmartLifecycle does those three plus autostartup, phases and adds a callback
on stop – gives a handle on the running thread to the application to end
gracefully.
- 17 -
Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
Polling – An architectural pattern (Better avoid)
•
The Spring Framework provides abstractions for asynchronous execution of
tasks with the TaskExecutor interface
•
The TaskExecutor interface has a single method execute(Runnable task)
•
The TaskExecutor gives Spring components an abstraction for thread pooling
where needed
•
Multiple pre-built implementations of TaskExecutor are included in Spring 3
which covers most of the patterns of task execution
- 18 -
Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
Task Execution using Executors
- 19 -
•
Important thing to
note is that the
execute method here
adds a Runnable
thread to the queue
and the TaskExecutor
uses its internal rules
to decide when the
task gets executed,
depending on the
Implementation class
we choose to inject.
•
We can easily
change a particular
behavior or pattern of
task execution by
changing the
Implementation
strategy.
Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
Task Execution – by configuration
•
@Async – adds an implicit TaskExecutor Support to a method
•
Spring internally uses a Dynamic Proxy for @Async annotated method so that
the method is actually executed by the implementation class of TaskExecutor
- 20 -
Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
Task Execution – @Async Annotation
•
The Spring Framework provides abstractions for asynchronous scheduling of
tasks with the TaskScheduler interface supported by Trigger and
TriggerContext interfaces
•
TaskScheduler supports recurring and cancelable tasks and they can be easily
controlled
- 21 -
Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
Task Scheduling
•
The Example below shows a POJO which can be turned into a business
service that can be executed as a Cron Job with basic configuration.
•
This Example uses Cron Expression “3/10 ****?” which is nothing but a 3
seconds offset and runs every 10 secs in a minute.
•
Other Possible Triggers are “fixed-delay” and “fixed-rate” which are configured
in milliseconds.
- 22 -
Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
Task Scheduling – by configuration
•
The @Scheduled annotation can be added to a method along with trigger
metadata
•
Start at Every Five seconds after completion of each run
•
Start at Every Five seconds after starting each run
•
Cron Job that executes on week days
- 23 -
Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
Task Scheduling – @Scheduled Annotation
•
Enterprise Application Integration (EAI) is an
application of technology defined as the
integration of data and services between
applications.
•
Spring Integration is an API from the creators
of the Spring Framework that's geared towards
Enterprise Application Integration (EAI).
•
Spring Integration offers improvement on
every integration solution that exists in Java
world.
•
Can be a powerful alternative to an ESB
•
Spring Integration addresses this concern –
“As the number of integration points increases
the applications will have to maintain and
manage many point to point channels of
communication and the convergence of these
channels at various end points create
“spaghetti architecture”.
Key Parts of Spring Integration : Message–Pay
- 24 -
Load–Header–Channel–Gateway
Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
Spring Integration – A bird-eye view
Spring Integration – A Simple Example
This example uses Spring Integration to process a book order and
appropriately route the message depending on if it's a pickup from the store or
if it should be delivered by post.
- 25 -
Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
•
- 26 Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
Spring Integration – A Simple Example – @Gateway
- 27 Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
Spring Integration – A Simple Example – @Router
- 28 Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
Spring Integration – A Simple Example – Bridge
- 29 Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
Spring Integration – A Simple Example – @Transformer
- 30 Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
Spring Integration – Outbound Channel Adapter
- 31 Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
Annotations – Dynamic Language Support – Spring EL
Channel Adapters and Message Gateways Supported
• JMS
• AMQP
• TCP
• UDP
• File/FTP/SFTP
• RMI
• RSS
• HTTP (REST)
• WS (SOAP/POX)
• JDBC
• XMPP
• Twitter
• Spring Events
- 32 -
Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
• Mail (POP3/IMAP/SMTP)
- 33 Karthik-Banda-Deloitte-Silicon-India-Java-Conference-Hyderabad_Final.pptx
Questions and Answers – bkarthik@deloitte.com
Copyright © 2011 Deloitte Development LLC. All rights reserved.