In this section we will give an overview of the HornetQ high level architecture.
HornetQ core is designed simply as set of Plain Old Java Objects (POJOs).
We've also designed it to have as few dependencies on external jars as possible. In fact, HornetQ core has zero dependencies on any jars other than the standard JDK classes!
This allows HornetQ to be easily embedded in your own project, or instantiated in any dependency injection framework such as JBoss Microcontainer, Spring or Google Guice.
A HornetQ server has its own high performance persistent journal, which it uses for message and other persistence.
Using a high performance journal allows outrageous persistence message performance, something not achievable when using a relational database for persistence.
HornetQ clients, potentially on different physical machines interact with the HornetQ server. HornetQ currently provides two APIs for messaging at the client side:
Core client API. This is a simple intuitive Java API that allows the full set of messaging functionality without some of the complexities of JMS.
JMS client API. The standard JMS API is available at the client side.
JMS semantics are implemented by a thin JMS facade layer on the client side.
The HornetQ server does not speak JMS and in fact does not know anything about JMS, it's a protocol agnostic messaging server designed to be used with multiple different protocols.
When a user uses the JMS API on the client side, all JMS interactions are translated into operations on the HornetQ core client API before being transferred over the wire using the HornetQ wire format.
The server always just deals with core API interactions.
A schematic illustrating this relationship is shown in figure 3.1 below:
Figure 3.1 shows two user applications interacting with a HornetQ server. User Application 1 is using the JMS API, while User Application 2 is using the core client API directly.
You can see from the diagram that the JMS API is implemented by a thin facade layer on the client side.
HornetQ core is designed as a set of simple POJOs so if you have an application that requires messaging functionality internally but you don't want to expose that as a HornetQ server you can directly instantiate and embed HornetQ servers in your own application.
For more information on embedding HornetQ, see Chapter 43, Embedding HornetQ.
HornetQ provides its own fully functional Java Connector Architecture (JCA) adaptor which enables it to be integrated easily into any JEE compliant application server or servlet engine.
JEE application servers provide Message Driven Beans (MDBs), which are a special type of Enterprise Java Beans (EJBs) that can process messages from sources such as JMS systems or mail systems.
Probably the most common use of an MDB is to consume messages from a JMS messaging system.
According to the JEE specification, a JEE application server uses a JCA adapter to integrate with a JMS messaging system so it can consume messages for MDBs.
However, the JCA adapter is not only used by the JEE application server for consuming messages via MDBs, it is also used when sending message to the JMS messaging system e.g. from inside an EJB or servlet.
When integrating with a JMS messaging system from inside a JEE application server it is always recommended that this is done via a JCA adaptor.
The application server's JCA service provides extra functionality such as connection pooling and automatic transaction enlistment, which are desirable when using messaging, say, from inside an EJB. It is possible to talk to a JMS messaging system directly from an EJB, MDB or servlet without going through a JCA adapter, but this is not recommended since you will not be able to take advantage of the JCA features, such as caching of JMS sessions, which can result in poor performance.
Figure 3.2 below shows a JEE application server integrating with a HornetQ server via the HornetQ JCA adaptor. Note that all communication between EJB sessions or entity beans and Message Driven beans go through the adaptor and not directly to HornetQ.
The large arrow with the prohibited sign shows an EJB session bean talking directly to the HornetQ server. This is not recommended as you'll most likely end up creating a new connection and session every time you want to interact from the EJB, which is an anti-pattern.
For more information on using the JCA adaptor, please see Chapter 33, Application Server Integration and Java EE.
HornetQ can also be deployed as a stand-alone server. This means a fully independent messaging server not dependent on a JEE application server.
The standard stand-alone messaging server configuration comprises a core messaging server, a JMS service and a JNDI service.
The role of the JMS Service is to deploy any JMS Queues, Topics and ConnectionFactory instances from any server side hornetq-jms.xml configuration files. It also provides a simple management API for creating and destroying Queues, Topics and ConnectionFactory instances which can be accessed via JMX or the connection. It is a separate service to the HornetQ core server, since the core server is JMS agnostic. If you don't want to deploy any JMS Queues, Topics and ConnectionFactory instances via server side XML configuration and don't require a JMS management API on the server side then you can disable this service.
We also include a JNDI server since JNDI is a common requirement when using JMS to lookup Queues, Topics and ConnectionFactory instances. If you do not require JNDI then this service can also be disabled. HornetQ allows you to programmatically create JMS and core objects directly on the client side as opposed to looking them up from JNDI, so a JNDI server is not always a requirement.
The stand-alone server configuration uses JBoss Microcontainer to instantiate and enforce dependencies between the components. JBoss Microcontainer is a very lightweight POJO bootstrapper.
The stand-alone server architecture is shown in figure 3.3 below:
For more information on server configuration files see Section 47.1, “Server Configuration”. $