One of the most compelling reasons to use the Spring Framework is the comprehensive transaction support. The Spring Framework provides a consistent abstraction for transaction management that delivers the following benefits: Provides a consistent programming model across different transaction APIs such as JTA, JDBC, Hibernate, JPA, and JDO. Supports declarative transaction management. Provides a simpler API for programmatic transaction management than a number of complex transaction APIs such as JTA. Integrates very well with Spring's various data access abstractions. This chapter is divided up into a number of sections, each detailing one of the value-adds or technologies of the Spring Framework's transaction support. The chapter closes up with some discussion of best practices surrounding transaction management (for example, choosing between declarative and programmatic transaction management). The first section, entitled Motivations, describes why one would want to use the Spring Framework's transaction abstraction as opposed to EJB CMT or driving transactions via a proprietary API such as Hibernate. The second section, entitled Key abstractions outlines the core classes in the Spring Framework's transaction support, as well as how to configure and obtain DataSource instances from a variety of sources. The third section, entitled Declarative transaction management, covers the Spring Framework's support for declarative transaction management. The fourth section, entitled Programmatic transaction management, covers the Spring Framework's support for programmatic (that is, explicitly coded) transaction management.

The Spring Framework's transaction management support significantly changes traditional thinking as to when a J2EE application requires an application server. In particular, you don't need an application server just to have declarative transactions via EJB. In fact, even if you have an application server with powerful JTA capabilities, you may well decide that the Spring Framework's declarative transactions offer more power and a much more productive programming model than EJB CMT. Typically you need an application server's JTA capability only if you need to enlist multiple transactional resources, and for many applications being able to handle transactions across multiple resources isn't a requirement. For example, many high-end applications use a single, highly scalable database (such as Oracle 9i RAC). Standalone transaction managers such as Atomikos Transactions and JOTM are other options. (Of course you may need other application server capabilities such as JMS and JCA.) The most important point is that with the Spring Framework you can choose when to scale your application up to a full-blown application server. Gone are the days when the only alternative to using EJB CMT or JTA was to write code using local transactions such as those on JDBC connections, and face a hefty rework if you ever needed that code to run within global, container-managed transactions. With the Spring Framework, only configuration needs to change so that your code doesn't have to. Traditionally, J2EE developers have had two choices for transaction management: global or local transactions. Global transactions are managed by the application server, using the Java Transaction API (JTA). Local transactions are resource-specific: the most common example would be a transaction associated with a JDBC connection. This choice has profound implications. For instance, global transactions provide the ability to work with multiple transactional resources (typically relational databases and message queues). With local transactions, the application server is not involved in transaction management and cannot help ensure correctness across multiple resources. (It is worth noting that most applications use a single transaction resource.) Global transactions have a significant downside, in that code needs to use JTA, and JTA is a cumbersome API to use (partly due to its exception model). Furthermore, a JTA UserTransaction normally needs to be sourced from JNDI: meaning that we need to use both JNDI and JTA to use JTA. Obviously all use of global transactions limits the reusability of application code, as JTA is normally only available in an application server environment. Previously, the preferred way to use global transactions was via EJB CMT (Container Managed Transaction): CMT is a form of declarative transaction management (as distinguished from programmatic transaction management). EJB CMT removes the need for transaction-related JNDI lookups - although of course the use of EJB itself necessitates the use of JNDI. It removes most of the need (although not entirely) to write Java code to control transactions. The significant downside is that CMT is tied to JTA and an application server environment. Also, it is only available if one chooses to implement business logic in EJBs, or at least behind a transactional EJB facade. The negatives around EJB in general are so great that this is not an attractive proposition, especially in the face of compelling alternatives for declarative transaction management. Local transactions may be easier to use, but have significant disadvantages: they cannot work across multiple transactional resources. For example, code that manages transactions using a JDBC connection cannot run within a global JTA transaction. Another downside is that local transactions tend to be invasive to the programming model. Spring resolves these problems. It enables application developers to use a consistent programming model in any environment. You write your code once, and it can benefit from different transaction management strategies in different environments. The Spring Framework provides both declarative and programmatic transaction management. Declarative transaction management is preferred by most users, and is recommended in most cases. With programmatic transaction management, developers work with the Spring Framework transaction abstraction, which can run over any underlying transaction infrastructure. With the preferred declarative model, developers typically write little or no code related to transaction management, and hence don't depend on the Spring Framework's transaction API (or indeed on any other transaction API).

The key to the Spring transaction abstraction is the notion of a transaction strategy. A transaction strategy is defined by the org.springframework.transaction.PlatformTransactionManager interface, shown below: This is primarily an SPI interface, although it can be used programmatically. Note that in keeping with the Spring Framework's philosophy, PlatformTransactionManager is an interface, and can thus be easily mocked or stubbed as necessary. Nor is it tied to a lookup strategy such as JNDI: PlatformTransactionManager implementations are defined like any other object (or bean) in the Spring Framework's IoC container. This benefit alone makes it a worthwhile abstraction even when working with JTA: transactional code can be tested much more easily than if it used JTA directly. Again in keeping with Spring's philosophy, the TransactionException that can be thrown by any of the PlatformTransactionManager interface's methods is unchecked (that is it extends the java.lang.RuntimeException class). Transaction infrastructure failures are almost invariably fatal. In rare cases where application code can actually recover from a transaction failure, the application developer can still choose to catch and handle TransactionException. The salient point is that developers are not forced to do so. The getTransaction(..) method returns a TransactionStatus object, depending on a TransactionDefinition parameter. The returned TransactionStatus might represent a new or existing transaction (if there were a matching transaction in the current call stack - with the implication being that (as with J2EE transaction contexts) a TransactionStatus is associated with a thread of execution). The TransactionDefinition interface specifies: Isolation: the degree of isolation this transaction has from the work of other transactions. For example, can this transaction see uncommitted writes from other transactions? Propagation: normally all code executed within a transaction scope will run in that transaction. However, there are several options specifying behavior if a transactional method is executed when a transaction context already exists: for example, simply continue running in the existing transaction (the common case); or suspending the existing transaction and creating a new transaction. Spring offers all of the transaction propagation options familiar from EJB CMT. (Some details regarding the semantics of transaction propagation in Spring can be found in the section entitled . Timeout: how long this transaction may run before timing out (and automatically being rolled back by the underlying transaction infrastructure). Read-only status: a read-only transaction does not modify any data. Read-only transactions can be a useful optimization in some cases (such as when using Hibernate). These settings reflect standard transactional concepts. If necessary, please refer to a resource discussing transaction isolation levels and other core transaction concepts because understanding such core concepts is essential to using the Spring Framework or indeed any other transaction management solution. The TransactionStatus interface provides a simple way for transactional code to control transaction execution and query transaction status. The concepts should be familiar, as they are common to all transaction APIs: Regardless of whether you opt for declarative or programmatic transaction management in Spring, defining the correct PlatformTransactionManager implementation is absolutely essential. In good Spring fashion, this important definition typically is made using via Dependency Injection. PlatformTransactionManager implementations normally require knowledge of the environment in which they work: JDBC, JTA, Hibernate, etc The following examples from the dataAccessContext-local.xml file from Spring's jPetStore sample application show how a local PlatformTransactionManager implementation can be defined. (This will work with plain JDBC.) We must define a JDBC DataSource, and then use the Spring DataSourceTransactionManager, giving it a reference to the DataSource. ]]> The related PlatformTransactionManager bean definition will look like this: ]]> If we use JTA in a J2EE container, as in the 'dataAccessContext-jta.xml' file from the same sample application, we use a container DataSource, obtained via JNDI, in conjunction with Spring's JtaTransactionManager. The JtaTransactionManager doesn't need to know about the DataSource, or any other specific resources, as it will use the container's global transaction management infrastructure. ]]><!-- other <bean/> definitions here -->]]> The above definition of the 'dataSource' bean uses the <jndi-lookup/> tag from the 'jee' namespace. For more information on schema-based configuration, see , and for more information on the <jee/> tags see the section entitled . We can also use Hibernate local transactions easily, as shown in the following examples from the Spring Framework's PetClinic sample application. In this case, we need to define a Hibernate LocalSessionFactoryBean, which application code will use to obtain Hibernate Session instances. The DataSource bean definition will be similar to the one shown previously (and thus is not shown). If the DataSource is managed by the JEE container it should be non-transactional as the Spring Framework, rather than the JEE container, will manage transactions. The 'txManager' bean in this case is of the HibernateTransactionManager type. In the same way as the DataSourceTransactionManager needs a reference to the DataSource, the HibernateTransactionManager needs a reference to the SessionFactory. org/springframework/samples/petclinic/hibernate/petclinic.hbm.xml hibernate.dialect=${hibernate.dialect} ]]> With Hibernate and JTA transactions, we can simply use the JtaTransactionManager as with JDBC or any other resource strategy. ]]> Note that this is identical to JTA configuration for any resource, as these are global transactions, which can enlist any transactional resource. In all these cases, application code will not need to change at all. We can change how transactions are managed merely by changing configuration, even if that change means moving from local to global transactions or vice versa.

It should now be clear how different transaction managers are created, and how they are linked to related resources which need to be synchronized to transactions (for example DataSourceTransactionManager to a JDBC DataSource, HibernateTransactionManager to a Hibernate SessionFactory, and so forth). There remains the question however of how the application code, directly or indirectly using a persistence API (such as JDBC, Hibernate, and JDO), ensures that these resources are obtained and handled properly in terms of proper creation/reuse/cleanup and trigger (optionally) transaction synchronization via the relevant PlatformTransactionManager.

The preferred approach is to use Spring's highest level persistence integration APIs. These do not replace the native APIs, but internally handle resource creation/reuse, cleanup, optional transaction synchronization of the resources and exception mapping so that user data access code doesn't have to worry about these concerns at all, but can concentrate purely on non-boilerplate persistence logic. Generally, the same template approach is used for all persistence APIs, with examples including the JdbcTemplate, HibernateTemplate, and JdoTemplate classes (detailed in subsequent chapters of this reference documentation.

At a lower level exist classes such as DataSourceUtils (for JDBC), SessionFactoryUtils (for Hibernate), PersistenceManagerFactoryUtils (for JDO), and so on. When it is preferable for application code to deal directly with the resource types of the native persistence APIs, these classes ensure that proper Spring Framework-managed instances are obtained, transactions are (optionally) synchronized, and exceptions which happen in the process are properly mapped to a consistent API. For example, in the case of JDBC, instead of the traditional JDBC approach of calling the getConnection() method on the DataSource, you would instead use Spring's org.springframework.jdbc.datasource.DataSourceUtils class as follows: If an existing transaction exists, and already has a connection synchronized (linked) to it, that instance will be returned. Otherwise, the method call will trigger the creation of a new connection, which will be (optionally) synchronized to any existing transaction, and made available for subsequent reuse in that same transaction. As mentioned, this has the added advantage that any SQLException will be wrapped in a Spring Framework CannotGetJdbcConnectionException - one of the Spring Framework's hierarchy of unchecked DataAccessExceptions. This gives you more information than can easily be obtained from the SQLException, and ensures portability across databases: even across different persistence technologies. It should be noted that this will also work fine without Spring transaction management (transaction synchronization is optional), so you can use it whether or not you are using Spring for transaction management. Of course, once you've used Spring's JDBC support or Hibernate support, you will generally prefer not to use DataSourceUtils or the other helper classes, because you'll be much happier working via the Spring abstraction than directly with the relevant APIs. For example, if you use the Spring JdbcTemplate or jdbc.object package to simplify your use of JDBC, correct connection retrieval happens behind the scenes and you won't need to write any special code.

At the very lowest level exists the TransactionAwareDataSourceProxy class. This is a proxy for a target DataSource, which wraps the target DataSource to add awareness of Spring-managed transactions. In this respect, it is similar to a transactional JNDI DataSource as provided by a J2EE server. It should almost never be necessary or desirable to use this class, except when existing code exists which must be called and passed a standard JDBC DataSource interface implementation. In that case, it's possible to still have this code be usable, but participating in Spring managed transactions. It is preferable to write your new code using the higher level abstractions mentioned above.

Most users of the Spring Framework choose declarative transaction management. It is the option with the least impact on application code, and hence is most consistent with the ideals of a non-invasive lightweight container. The Spring Framework's declarative transaction management is made possible with Spring AOP, although, as the transactional aspects code comes with the Spring Framework distribution and may be used in a boilerplate fashion, AOP concepts do not generally have to be understood to make effective use of this code. It may be helpful to begin by considering EJB CMT and explaining the similarities and differences with the Spring Framework's declarative transaction management. The basic approach is similar: it is possible to specify transaction behavior (or lack of it) down to individual method level. It is possible to make a setRollbackOnly() call within a transaction context if necessary. The differences are: Unlike EJB CMT, which is tied to JTA, the Spring Framework's declarative transaction management works in any environment. It can work with JDBC, JDO, Hibernate or other transactions under the covers, with configuration changes only. The Spring Framework enables declarative transaction management to be applied to any class, not merely special classes such as EJBs. The Spring Framework offers declarative rollback rules: this is a feature with no EJB equivalent. Both programmatic and declarative support for rollback rules is provided. The Spring Framework gives you an opportunity to customize transactional behavior, using AOP. For example, if you want to insert custom behavior in the case of transaction rollback, you can. You can also add arbitrary advice, along with the transactional advice. With EJB CMT, you have no way to influence the container's transaction management other than setRollbackOnly(). The Spring Framework does not support propagation of transaction contexts across remote calls, as do high-end application servers. If you need this feature, we recommend that you use EJB. However, consider carefully before using such a feature, because normally, one does not want transactions to span remote calls. Declarative transaction configuration in versions of Spring 2.0 and above differs considerably from previous versions of Spring. The main difference is that there is no longer any need to configure TransactionProxyFactoryBean beans. The old, pre-Spring 2.0 configuration style is still 100% valid configuration; think of the new <tx:tags/> as simply defining TransactionProxyFactoryBean beans on your behalf. The concept of rollback rules is important: they enable us to specify which exceptions (and throwables) should cause automatic roll back. We specify this declaratively, in configuration, not in Java code. So, while we can still call setRollbackOnly()on the TransactionStatus object to roll the current transaction back programmatically, most often we can specify a rule that MyApplicationException must always result in rollback. This has the significant advantage that business objects don't need to depend on the transaction infrastructure. For example, they typically don't need to import any Spring APIs, transaction or other. While the EJB default behavior is for the EJB container to automatically roll back the transaction on a system exception (usually a runtime exception), EJB CMT does not roll back the transaction automatically on an application exception (that is, a checked exception other than java.rmi.RemoteException). While the Spring default behavior for declarative transaction management follows EJB convention (roll back is automatic only on unchecked exceptions), it is often useful to customize this.

The aim of this section is to dispel the mystique that is sometimes associated with the use of declarative transactions. It is all very well for this reference documentation simply to tell you to annotate your classes with the @Transactional annotation, add the line ('<tx:annotation-driven/>') to your configuration, and then expect you to understand how it all works. This section will explain the inner workings of the Spring Framework's declarative transaction infrastructure to help you navigate your way back upstream to calmer waters in the event of transaction-related issues. The most important concepts to grasp with regard to the Spring Framework's declarative transaction support are that this support is enabled via AOP proxies, and that the transactional advice is driven by metadata (currently XML- or annotation-based). The combination of AOP with transactional metadata yields an AOP proxy that uses a TransactionInterceptor in conjunction with an appropriate PlatformTransactionManager implementation to drive transactions around method invocations. Although knowledge of Spring AOP is not required to use Spring's declarative transaction support, it can help. Spring AOP is thoroughly covered in the chapter entitled . Conceptually, calling a method on a transactional proxy looks like this...

Consider the following interface, and its attendant implementation. (The intent is to convey the concepts, and using the rote Foo and Bar tropes means that you can concentrate on the transaction usage and not have to worry about the domain model.) // the service interface that we want to make transactional // an implementation of the above interface (For the purposes of this example, the fact that the DefaultFooService class throws UnsupportedOperationException instances in the body of each implemented method is good; it will allow us to see transactions being created and then rolled back in response to the UnsupportedOperationException instance being thrown.) Let's assume that the first two methods of the FooService interface (getFoo(String) and getFoo(String, String)) have to execute in the context of a transaction with read-only semantics, and that the other methods (insertFoo(Foo) and updateFoo(Foo)) have to execute in the context of a transaction with read-write semantics. Don't worry about taking the following configuration in all at once; everything will be explained in detail in the next few paragraphs. <!-- from the file 'context.xml' --> xmlns:tx="http://www.springframework.org/schema/tx"http://www.springframework.org/schema/tx http://www.springframework.org/schema/tx/spring-tx-2.5.xsd ]]><!-- this is the service object that we want to make transactional --> ]]><!-- the transactional advice (what 'happens'; see the <aop:advisor/> bean below) --> ]]><!-- the transactional semantics... --> ]]><!-- all methods starting with 'get' are read-only --> ]]><!-- other methods use the default transaction settings (see below) --> ]]><!-- ensure that the above transactional advice runs for any execution of an operation defined by the FooService interface --> ]]><!-- don't forget the DataSource --> ]]><!-- similarly, don't forget the PlatformTransactionManager --> ]]><!-- other <bean/> definitions here -->]]> Let's pick apart the above configuration. We have a service object (the 'fooService' bean) that we want to make transactional. The transaction semantics that we want to apply are encapsulated in the <tx:advice/> definition. The <tx:advice/> definition reads as ... all methods on starting with 'get' are to execute in the context of a read-only transaction, and all other methods are to execute with the default transaction semantics. The 'transaction-manager' attribute of the <tx:advice/> tag is set to the name of the PlatformTransactionManager bean that is going to actually drive the transactions (in this case the 'txManager' bean). You can actually omit the 'transaction-manager' attribute in the transactional advice (<tx:advice/>) if the bean name of the PlatformTransactionManager that you want to wire in has the name 'transactionManager'. If the PlatformTransactionManager bean that you want to wire in has any other name, then you have to be explicit and use the 'transaction-manager' attribute as in the example above. The <aop:config/> definition ensures that the transactional advice defined by the 'txAdvice' bean actually executes at the appropriate points in the program. First we define a pointcut that matches the execution of any operation defined in the FooService interface ('fooServiceOperation'). Then we associate the pointcut with the 'txAdvice' using an advisor. The result indicates that at the execution of a 'fooServiceOperation', the advice defined by 'txAdvice' will be run. The expression defined within the <aop:pointcut/> element is an AspectJ pointcut expression; see the chapter entitled for more details on pointcut expressions in Spring 2.0. A common requirement is to make an entire service layer transactional. The best way to do this is simply to change the pointcut expression to match any operation in your service layer. For example: ]]> (This example assumes that all your service interfaces are defined in the 'x.y.service' package; see the chapter entitled for more details.) Now that we've analyzed the configuration, you may be asking yourself, Okay... but what does all this configuration actually do?. The above configuration is going to effect the creation of a transactional proxy around the object that is created from the 'fooService' bean definition. The proxy will be configured with the transactional advice, so that when an appropriate method is invoked on the proxy, a transaction may be started, suspended, be marked as read-only, etc., depending on the transaction configuration associated with that method. Consider the following program that test drives the above configuration. The output from running the above program will look something like this. (Please note that the Log4J output and the stacktrace from the UnsupportedOperationException thrown by the insertFoo(..) method of the DefaultFooService class have been truncated in the interest of clarity.) <!-- the Spring container is starting up... --><!-- the DefaultFooService is actually proxied --><!-- ... the insertFoo(..) method is now being invoked on the proxy --><!-- the transactional advice kicks in here... --><!-- the insertFoo(..) method from DefaultFooService throws an exception... --><!-- and the transaction is rolled back (by default, RuntimeException instances cause rollback) --><!-- AOP infrastructure stack trace elements removed for clarity -->

The previous section outlined the basics of how to specify the transactional settings for the classes, typically service layer classes, in your application in a declarative fashion. This section describes how you can control the rollback of transactions in a simple declarative fashion. The recommended way to indicate to the Spring Framework's transaction infrastructure that a transaction's work is to be rolled back is to throw an Exception from code that is currently executing in the context of a transaction. The Spring Framework's transaction infrastructure code will catch any unhandled Exception as it bubbles up the call stack, and will mark the transaction for rollback. Note however that the Spring Framework's transaction infrastructure code will, by default, only mark a transaction for rollback in the case of runtime, unchecked exceptions; that is, when the thrown exception is an instance or subclass of RuntimeException. (Errors will also - by default - result in a rollback.) Checked exceptions that are thrown from a transactional method will not result in the transaction being rolled back. Exactly which Exception types mark a transaction for rollback can be configured. Find below a snippet of XML configuration that demonstrates how one would configure rollback for a checked, application-specific Exception type. rollback-for="NoProductInStockException" ]]> It is also possible to specify 'no rollback rules', for those times when you do not want a transaction to be marked for rollback when an exception is thrown. In the example configuration below, we effectively are telling the Spring Framework's transaction infrastructure to commit the attendant transaction even in the face of an unhandled InstrumentNotFoundException. no-rollback-for="InstrumentNotFoundException" ]]> When the Spring Framework's transaction infrastructure has caught an exception and is consulting any configured rollback rules to determine whether or not to mark the transaction for rollback, the strongest matching rule wins. So in the case of the following configuration, any exception other than an InstrumentNotFoundException would result in the attendant transaction being marked for rollback. ]]> The second way to indicate that a rollback is required is to do so programmatically. Although very simple, this way is quite invasive, and tightly couples your code to the Spring Framework's transaction infrastructure, as can be seen below: // some business logic...// trigger rollback programmatically You are strongly encouraged to use the declarative approach to rollback if at all possible. Programmatic rollback is available should you absolutely need it, but its usage flies in the face of achieving a nice, clean POJO-based architecture.

Consider the scenario where you have a number of service layer objects, and you want to apply totally different transactional configuration to each of them. This is achieved by defining distinct <aop:advisor/> elements with differing 'pointcut' and 'advice-ref' attribute values. Let's assume that all of your service layer classes are defined in a root 'x.y.service' package. To make all beans that are instances of classes defined in that package (or in subpackages) and that have names ending in 'Service' have the default transactional configuration, you would write the following: ]]><!-- these two beans will be transactional... --> ]]><!-- ... and these two beans won't --> ]]><!-- (not in the right package) --> ]]><!-- (doesn't end in 'Service') --> ]]><!-- other transaction infrastructure beans such as a PlatformTransactionManager omitted... -->]]> Find below an example of configuring two distinct beans with totally different transactional settings. ]]><!-- this bean will be transactional (see the 'defaultServiceOperation' pointcut) --> ]]><!-- this bean will also be transactional, but with totally different transactional settings --> ]]><!-- other transaction infrastructure beans such as a PlatformTransactionManager omitted... -->]]>

This section summarises the various transactional settings that can be specified using the <tx:advice/> tag. The default <tx:advice/> settings are: The propagation setting is REQUIRED The isolation level is DEFAULT The transaction is read/write The transaction timeout defaults to the default timeout of the underlying transaction system, or or none if timeouts are not supported Any RuntimeException will trigger rollback, and any checked Exception will not These default settings can be changed; the various attributes of the <tx:method/> tags that are nested within <tx:advice/> and <tx:attributes/> tags are summarized below: Attribute Required? Default Description name Yes The method name(s) with which the transaction attributes are to be associated. The wildcard (*) character can be used to associate the same transaction attribute settings with a number of methods; for example, 'get*', 'handle*', 'on*Event', and so forth. propagation No REQUIRED The transaction propagation behavior isolation No DEFAULT The transaction isolation level timeout No -1 The transaction timeout value (in seconds) read-only No false Is this transaction read-only? rollback-for No The Exception(s) that will trigger rollback; comma-delimited. For example, 'com.foo.MyBusinessException,ServletException' no-rollback-for No The Exception(s) that will not trigger rollback; comma-delimited. For example, 'com.foo.MyBusinessException,ServletException'

The functionality offered by the @Transactional annotation and the support classes is only available to you if you are using at least Java 5 (Tiger). In addition to the XML-based declarative approach to transaction configuration, you can also use an annotation-based approach to transaction configuration. Declaring transaction semantics directly in the Java source code puts the declarations much closer to the affected code, and there is generally not much danger of undue coupling, since code that is meant to be used transactionally is almost always deployed that way anyway. The ease-of-use afforded by the use of the @Transactional annotation is best illustrated with an example, after which all of the details will be explained. Consider the following class definition: // the service class that we want to make transactional@Transactional When the above POJO is defined as a bean in a Spring IoC container, the bean instance can be made transactional by adding merely one line of XML configuration, like so: <!-- from the file 'context.xml' --> ]]><!-- this is the service object that we want to make transactional --> ]]><!-- enable the configuration of transactional behavior based on annotations -->]]><!-- a PlatformTransactionManager is still required --> ]]><!-- (this dependency is defined somewhere else) --> ]]><!-- other <bean/> definitions here -->]]> You can actually omit the 'transaction-manager' attribute in the <tx:annotation-driven/> tag if the bean name of the PlatformTransactionManager that you want to wire in has the name 'transactionManager'. If the PlatformTransactionManager bean that you want to dependency inject has any other name, then you have to be explicit and use the 'transaction-manager' attribute as in the example above. When using proxies, the @Transactional annotation should only be applied to methods with public visibility. If you do annotate protected, private or package-visible methods with the @Transactional annotation, no error will be raised, but the annotated method will not exhibit the configured transactional settings. Consider the use of AspectJ (see below) if you need to annotate non-public methods. The @Transactional annotation may be placed before an interface definition, a method on an interface, a class definition, or a public method on a class. However, please note that the mere presence of the @Transactional annotation is not enough to actually turn on the transactional behavior - the @Transactional annotation is simply metadata that can be consumed by something that is @Transactional-aware and that can use the metadata to configure the appropriate beans with transactional behavior. In the case of the above example, it is the presence of the <tx:annotation-driven/> element that switches on the transactional behavior. The Spring team's recommendation is that you only annotate concrete classes with the @Transactional annotation, as opposed to annotating interfaces. You certainly can place the @Transactional annotation on an interface (or an interface method), but this will only work as you would expect it to if you are using interface-based proxies. The fact that annotations are not inherited means that if you are using class-based proxies (proxy-target-class="true") or the weaving-based aspect (mode="aspectj") then the transaction settings will not be recognised by the proxying/weaving infrastructure and the object will not be wrapped in a transactional proxy (which would be decidedly bad). So please do take the Spring team's advice and only annotate concrete classes (and the methods of concrete classes) with the @Transactional annotation. Note: In proxy mode (which is the default), only 'external' method calls coming in through the proxy will be intercepted. This means that 'self-invocation', i.e. a method within the target object calling some other method of the target object, won't lead to an actual transaction at runtime even if the invoked method is marked with @Transactional! Consider the use of AspectJ mode (see below) if you expect self-invocations to be wrapped with transactions as well. In this case, there won't be a proxy in the first place; instead, the target class will be 'weaved' (i.e. its byte code will be modified) in order to turn @Transactional into runtime behavior on any kind of method. Attribute Default Description transaction-manager transactionManager The name of transaction manager to use. Only required if the name of the transaction manager is not transactionManager, as in the example above. mode proxy The default mode "proxy" will process annotated beans to be proxied using Spring's AOP framework (following proxy semantics, as discussed above, applying to method calls coming in through the proxy only). The alternative mode "aspectj" will instead weave the affected classes with Spring's AspectJ transaction aspect (modifying the target class byte code in order to apply to any kind of method call). AspectJ weaving requires spring-aspects.jar on the classpath as well as load-time weaving (or compile-time weaving) enabled. (See the section entitled for details on how to set up load-time weaving.) proxy-target-class false Applies to proxy mode only. Controls what type of transactional proxies are created for classes annotated with the @Transactional annotation. If "proxy-target-class" attribute is set to "true", then class-based proxies will be created. If "proxy-target-class" is "false" or if the attribute is omitted, then standard JDK interface-based proxies will be created. (See the section entitled for a detailed examination of the different proxy types.) order Ordered.LOWEST_PRECEDENCE Defines the order of the transaction advice that will be applied to beans annotated with @Transactional. More on the rules related to ordering of AOP advice can be found in the AOP chapter (see section ). Note that not specifying any ordering will leave the decision as to what order advice is run in to the AOP subsystem. The "proxy-target-class" attribute on the <tx:annotation-driven/> element controls what type of transactional proxies are created for classes annotated with the @Transactional annotation. If "proxy-target-class" attribute is set to "true", then class-based proxies will be created. If "proxy-target-class" is "false" or if the attribute is omitted, then standard JDK interface-based proxies will be created. (See the section entitled for a detailed examination of the different proxy types.) Note that <tx:annotation-driven/> only looks for @Transactional on beans in the same application context it is defined in. This means that, if you put <tx:annotation-driven/> in a WebApplicationContext for a DispatcherServlet, it only checks for @Transactional beans in your controllers, and not your services. See for more information. The most derived location takes precedence when evaluating the transactional settings for a method. In the case of the following example, the DefaultFooService class is annotated at the class level with the settings for a read-only transaction, but the @Transactional annotation on the updateFoo(Foo) method in the same class takes precedence over the transactional settings defined at the class level. // do something// these settings have precedence for this method// do something

The @Transactional annotation is metadata that specifies that an interface, class, or method must have transactional semantics; for example, start a brand new read-only transaction when this method is invoked, suspending any existing transaction. The default @Transactional settings are: The propagation setting is PROPAGATION_REQUIRED The isolation level is ISOLATION_DEFAULT The transaction is read/write The transaction timeout defaults to the default timeout of the underlying transaction system, or or none if timeouts are not supported Any RuntimeException will trigger rollback, and any checked Exception will not These default settings can be changed; the various properties of the @Transactional annotation are summarized in the following table: Property Type Description propagation enum: Propagation optional propagation setting isolation enum: Isolation optional isolation level readOnly boolean read/write vs. read-only transaction timeout int (in seconds granularity) the transaction timeout rollbackFor an array of Class objects, which must be derived from Throwable an optional array of exception classes which must cause rollback rollbackForClassname an array of class names. Classes must be derived from Throwable an optional array of names of exception classes that must cause rollback noRollbackFor an array of Class objects, which must be derived from Throwable an optional array of exception classes that must not cause rollback. noRollbackForClassname an array of String class names, which must be derived from Throwable an optional array of names of exception classes that must not cause rollback Currently it is not possible to have explicit control over the name of a transaction, where 'name' means the transaction name that will be shown in a transaction monitor, if applicable (for example, WebLogic's transaction monitor), and in logging output. For declarative transactions, the transaction name is always the fully-qualified class name + "." + method name of the transactionally-advised class. For example, if the handlePayment(..) method of the BusinessService class started a transaction, the name of the transaction would be: com.foo.BusinessService.handlePayment.

Please note that this section of the Spring reference documentation is not an introduction to transaction propagation proper; rather it details some of the semantics regarding transaction propagation in Spring. In the case of Spring-managed transactions, please be aware of the difference between physical and logical transactions, and how the propagation setting applies to this difference.

PROPAGATION_REQUIRED When the propagation setting is PROPAGATION_REQUIRED, a logical transaction scope is created for each method that it gets applied to. Each such logical transaction scope can individually decide on rollback-only status, with an outer transaction scope being logically independent from the inner transaction scope. Of course, in case of standard PROPAGATION_REQUIRED behavior, they will be mapped to the same physical transaction. So a rollback-only marker set in the inner transaction scope does affect the outer transactions chance to actually commit (as you would expect it to). However, in the case where an inner transaction scopes sets the rollback-only marker, the outer transaction itself has not decided on the rollback itself, and so the rollback (silently triggered by the inner transaction scope) is unexpected: a corresponding UnexpectedRollbackException will be thrown at that point. This is expected behavior so that the caller of a transaction can never be misled to assume that a commit was performed when it really was not. So if an inner transaction (that the outer caller is not aware of) silently marks a transaction as rollback-only, the outer caller would still innocently call commit - and needs to receive an UnexpectedRollbackException to indicate clearly that a rollback was performed instead.

PROPAGATION_REQUIRES_NEW PROPAGATION_REQUIRES_NEW, in contrast, uses a completely independent transaction for each affected transaction scope. In that case, the underlying physical transactions will be different and hence can commit or rollback independently, with an outer transaction not affected by an inner transaction's rollback status.

PROPAGATION_NESTED is different again in that it uses a single physical transaction with multiple savepoints that it can roll back to. Such partial rollbacks allow an inner transaction scope to trigger a rollback for its scope, with the outer transaction being able to continue the physical transaction despite some operations having been rolled back. This is typically mapped onto JDBC savepoints, so will only work with JDBC resource transactions (see Spring's DataSourceTransactionManager).

Consider the situation where you would like to execute both transactional and (to keep things simple) some basic profiling advice. How do you effect this in the context of using <tx:annotation-driven/>? What we want to see when we invoke the updateFoo(Foo) method is: the configured profiling aspect starting up, then the transactional advice executing, then the method on the advised object executing then the transaction committing (we'll assume a sunny day scenario here), and then finally the profiling aspect reporting (somehow) exactly how long the whole transactional method invocation took This chapter is not concerned with explaining AOP in any great detail (except as it applies to transactions). Please see the chapter entitled for detailed coverage of the various bits and pieces of the following AOP configuration (and AOP in general). Here is the code for a simple profiling aspect. The ordering of advice is controlled via the Ordered interface. For full details on advice ordering, see . // allows us to control the ordering of advice// this method is the around advice ]]><!-- this is the aspect --> ]]><!-- execute before the transactional advice (hence the lower order number) -->value="1" order="200" ]]><!-- this advice will execute around the transactional advice --> ]]> The result of the above configuration will be a 'fooService' bean that has profiling and transactional aspects applied to it in that order. The configuration of any number of additional aspects is effected in a similar fashion. Finally, find below some example configuration for effecting the same setup as above, but using the purely XML declarative approach. ]]><!-- the profiling advice --> ]]><!-- execute before the transactional advice (hence the lower order number) --><property name="order" value="1 ]]><!-- will execute after the profiling advice (c.f. the order attribute) -->order="2"/> <!-- order value is higher than the profiling aspect --> ]]><!-- other <bean/> definitions such as a DataSource and a PlatformTransactionManager here -->]]> The result of the above configuration will be a 'fooService' bean that has profiling and transactional aspects applied to it in that order. If we wanted the profiling advice to execute after the transactional advice on the way in, and before the transactional advice on the way out, then we would simply swap the value of the profiling aspect bean's 'order' property such that it was higher than the transactional advice's order value. The configuration of any number of additional aspects is achieved in a similar fashion.

It is also possible to use the Spring Framework's @Transactional support outside of a Spring container by means of an AspectJ aspect. To use this support you must first annotate your classes (and optionally your classes' methods with the @Transactional annotation, and then you must link (weave) your application with the org.springframework.transaction.aspectj.AnnotationTransactionAspect defined in the spring-aspects.jar file. The aspect must also be configured with a transaction manager. You could of course use the Spring Framework's IoC container to take care of dependency injecting the aspect. The simplest way to configure the transaction management aspect is to use the '<tx:annotation-driven/>' element and specify the mode attribute to asepctj as described in . Since we're focusing here on applications running outside of a Spring container, we'll show you how to do it programmatically. Prior to continuing, you may well want to read the previous sections entitled and respectively. // construct an appropriate transaction manager // configure the AnnotationTransactionAspect to use it; this must be done before executing any transactional methods When using this aspect, you must annotate the implementation class (and/or methods within that class), not the interface (if any) that the class implements. AspectJ follows Java's rule that annotations on interfaces are not inherited. The @Transactional annotation on a class specifies the default transaction semantics for the execution of any method in the class. The @Transactional annotation on a method within the class overrides the default transaction semantics given by the class annotation (if present). Any method may be annotated, regardless of visibility. To weave your applications with the AnnotationTransactionAspect you must either build your application with AspectJ (see the AspectJ Development Guide) or use load-time weaving. See the section entitled for a discussion of load-time weaving with AspectJ.

The Spring Framework provides two means of programmatic transaction management: Using the TransactionTemplate. Using a PlatformTransactionManager implementation directly. If you are going to use programmatic transaction management, the Spring team generally recommends using the TransactionTemplate. The second approach is similar to using the JTA UserTransaction API (although exception handling is less cumbersome).

The TransactionTemplate adopts the same approach as other Spring templates such as the JdbcTemplate. It uses a callback approach, to free application code from having to do the boilerplate acquisition and release of transactional resources, and results in code that is intention driven, in that the code that is written focuses solely on what the developer wants to do. As you will immediately see in the examples that follow, using the TransactionTemplate absolutely couples you to Spring's transaction infrastructure and APIs. Whether or not programmatic transaction management is suitable for your development needs is a decision that you will have to make yourself. Application code that must execute in a transactional context, and that will use the TransactionTemplate explicitly, looks like this. You, as an application developer, will write a TransactionCallback implementation (typically expressed as an anonymous inner class) that will contain all of the code that you need to have execute in the context of a transaction. You will then pass an instance of your custom TransactionCallback to the execute(..) method exposed on the TransactionTemplate. // single TransactionTemplate shared amongst all methods in this instance// use constructor-injection to supply the PlatformTransactionManager// the code in this method executes in a transactional context If there is no return value, use the convenient TransactionCallbackWithoutResult class via an anonymous class like so: TransactionCallbackWithoutResult Code within the callback can roll the transaction back by calling the setRollbackOnly() method on the supplied TransactionStatus object. status.setRollbackOnly();

Transaction settings such as the propagation mode, the isolation level, the timeout, and so forth can be set on the TransactionTemplate either programmatically or in configuration. TransactionTemplate instances by default have the default transactional settings. Find below an example of programmatically customizing the transactional settings for a specific TransactionTemplate. // the transaction settings can be set here explicitly if so desired// 30 seconds// and so forth... Find below an example of defining a TransactionTemplate with some custom transactional settings, using Spring XML configuration. The 'sharedTransactionTemplate' can then be injected into as many services as are required. "]]>

Finally, instances of the TransactionTemplate class are threadsafe, in that instances do not maintain any conversational state. TransactionTemplate instances do however maintain configuration state, so while a number of classes may choose to share a single instance of a TransactionTemplate, if a class needed to use a TransactionTemplate with different settings (for example, a different isolation level), then two distinct TransactionTemplate instances would need to be created and used.

You can also use the org.springframework.transaction.PlatformTransactionManager directly to manage your transaction. Simply pass the implementation of the PlatformTransactionManager you're using to your bean via a bean reference. Then, using the TransactionDefinition and TransactionStatus objects you can initiate transactions, rollback and commit. // explicitly setting the transaction name is something that can only be done programmatically// execute your business logic here

Programmatic transaction management is usually a good idea only if you have a small number of transactional operations. For example, if you have a web application that require transactions only for certain update operations, you may not want to set up transactional proxies using Spring or any other technology. In this case, using the TransactionTemplate may be a good approach. Being able to set the transaction name explicitly is also something that can only be done using the programmatic approach to transaction management. On the other hand, if your application has numerous transactional operations, declarative transaction management is usually worthwhile. It keeps transaction management out of business logic, and is not difficult to configure. When using the Spring Framework, rather than EJB CMT, the configuration cost of declarative transaction management is greatly reduced.

Spring's transaction abstraction generally is application server agnostic. Additionally, Spring's JtaTransactionManager class, which can optionally perform a JNDI lookup for the JTA UserTransaction and TransactionManager objects, autodetects the location for the latter object, which varies by application server. Having access to the JTA TransactionManager allows for enhanced transaction semantics, in particular supporting transaction suspension. Please see the JtaTransactionManager Javadocs for details. Spring's JtaTransactionManager is the standard choice when running on J2EE application servers, known to work on all common servers. Its advanced functionality such as transaction suspension is known to work on many servers as well - including GlassFish, JBoss, Geronimo and Oracle OC4J - without any special configuration required. However, for fully supported transaction suspension and further advanced integration, Spring ships special adapters for IBM WebSphere and BEA WebLogic and also for Oracle OC4J. We'll discuss these adapters in the following sections. For standard scenarios, including WebLogic, WebSphere and OC4J, consider using the convenient '<tx:jta-transaction-manager/>' configuration element. This will automatically detect the underlying server and choose the best transaction manager available for the platform. This means that you won't have to configure server-specific adapter classes (as discussed in the following sections) explicitly; they will rather be chosen automatically, with the standard JtaTransactionManager as default fallback.

On WebSphere 6.0 and above, the recommended Spring JTA transaction manager to use is WebSphereUowTransactionManager. This special adapter leverages IBM's UOWManager API which is available in WebSphere Application Server 6.0.2.19 or above and 6.1.0.9 or above. With this adapter, Spring-driven transaction suspension (suspend/resume as initiated by PROPAGATION_REQUIRES_NEW) is officially supported by IBM! In a WebSphere 5.1 environment, you may wish to use Spring's WebSphereTransactionManagerFactoryBean class. This is a factory bean which retrieves the JTA TransactionManager in a WebSphere environment, which is done via WebSphere's static access methods. Once the JTA TransactionManager instance has been obtained via this factory bean, Spring's JtaTransactionManager may be configured with a reference to it, for enhanced transaction semantics over the use of only the JTA UserTransaction object. Please see the Javadocs for full details. Note that WebSphereTransactionManagerFactoryBean usage is known to work on WAS 5.1 and 6.0 but is not officially supported by IBM. Prefer WebSphereUowTransactionManager when running on WAS 6.0 or higher (see above).

On WebLogic 8.1 or above, you will generally prefer to use the WebLogicJtaTransactionManager instead of the stock JtaTransactionManager class. This special WebLogic-specific subclass of the normal JtaTransactionManager supports the full power of Spring's transaction definitions in a WebLogic-managed transaction environment, beyond standard JTA semantics: Features include transaction names, per-transaction isolation levels, and proper resuming of transactions in all cases.

Spring ships a special adapter class for OC4J 10.1.3 or above: OC4JJtaTransactionManager. This is analogous to the WebLogicJtaTransactionManager class discussed in the previous section, providing similar value-adds on OC4J: transaction names and per-transaction isolation levels. Note that the full JTA functionality, including transaction suspension, works fine with Spring's JtaTransactionManager on OC4J as well. The special OC4JJtaTransactionManager adapter simply provides value-adds beyond standard JTA.

You should take care to use the correct PlatformTransactionManager implementation for their requirements. Used properly, the Spring Framework merely provides a straightforward and portable abstraction. If you are using global transactions, you must use the org.springframework.transaction.jta.JtaTransactionManager class (or an application server-specific subclass of it) for all your transactional operations. Otherwise the transaction infrastructure will attempt to perform local transactions on resources such as container DataSource instances. Such local transactions do not make sense, and a good application server will treat them as errors.

Find below links to further resources about the Spring Framework's transaction support. Java Transaction Design Strategies is a book available from InfoQ that provides a well-paced introduction to transactions in Java. It also includes side-by-side examples of how to configure and use transactions using both the Spring Framework and EJB3.