RETRIEVING AN OBJECT

Abstract

A method of locating an object across a network comprises locating the copy of object 5 that is the best copy available through either JNDI object location and webservice location. In one advantageous embodiment, the method initially comprises attempting to locate a local copy of the object (step 8). If a local copy of the object is not located, the method 10 then attempts (step 10) to locate a remote copy of the object. If a remote copy of the object is not located, the method then attempts (step 12) to locate a remote web service object.

Description

TECHNICAL FIELD

The invention relates to a method of retrieving an object across a network, and in particular to a method that bridges JNDI (Java Naming and Directory Interface) lookup and web service location to retrieve the copy of the object that is the best copy available through either JNDI look-up or web service location.

BACKGROUND

The Java Naming and Directory Interface is an interface that allows clients to look-up an object (a data item) via a name. The Java Naming and Directory Interface essentially provides a mechanism for binding an object to a name, and a directory lookup interface that allows a client who wises to retrieve an object to enter a query.

Service locators that use JNDI for location of a service object are known, for example see http://java.sun.com/blueprints/corej2eepatterns/Patterns/ServiceLocator.html. This addresses the problem that different clients use JNDI in different ways, for example EJB (Enterprise JavaBeans) clients use JNDI to look up EJBHome objects by using the enterprise bean's registered JNDI name, where JMS (Java Messaging Service) clients need to look up JMS components by using the JNDI names registered for JMS components. It proposes a “Service Locator object” that can abstract all JNDI usage and hide the complexities of initial context creation, EJB home object lookup, and EJB object re-creation. Multiple clients can reuse the Service Locator object to reduce code complexity and provide a single point of control.

However, with EJB 3.0 (Enterprise JavaBeans 3.0) and IOC (Inversion of Control), the usefulness of the existing Service Locator pattern is reduced as beans are injected with annotations instead of being looked up from the JNDI. Furthermore, it covers only abstraction of the JNDI usage and initial context creation.

SUMMARY

A first aspect of the invention provides a method of retrieving an object. The method comprises retrieving the object that is the best available through either JNDI object location and webservice location. The term “best available object” denotes the copy of the object that can retrieved most quickly.

The invention thus bridges the JNDI object location and webservice location, and provides increased probability of successful retrieval of an object compared to use of JNDI alone while allowing more efficient retrieval of an object than if web service location alone is used.

Web service location using EJB 3 is described at, for example, the following, to which attention is directed: http://www.theregister.co.uk/2007/01/23/ejb_web_services/

The invention may comprise, in response to a request to retrieve an object, initially carrying out a JNDI search for the object. If the JNDI search locates a copy of the object the object is retrieved, but if the JNDI search does not locate a copy of the object, web service location is used to try to locate a remote web service object. In general, web service location is at significantly slower in retrieving an object than a JNDI search. Thus, attempting to locate the object with a JNDI search and using a web service location only if the JNDI fails to locate a copy of the object should reduce the time taken to locate a copy of the object.

Carrying out the JNDI search may comprise initially attempting to locate a local copy of the object. If a local copy of the object is located this is retrieved, but if a local copy is not located, the method then attempts to locate a remote a copy of the object. Invoking a request across a local JNDI call is approximately 5-6 times faster than invoking the request across a remote call, so that initially attempting to locate a local copy of the object should reduce the time taken to locate a copy of the object.

The method may comprise attempting to locate a requested object in a cache before attempting to retrieve the object.

The method may comprise caching a retrieved object. Caching a retrieved object means that the search cost is incurred only when the first request for an object is received, as subsequent requests for that object can be met by returning the cached copy of the object (assuming that it is still up-to-date).

Other aspects of the invention provide a corresponding apparatus and a corresponding storage medium.

BRIEF DESCRIPTION OF THE DRAWINGS

Preferred embodiments of the present invention will now be described by way of illustrative example with reference to the accompanying drawings in which:

FIG. 1 is a block flow diagram showing principal step of a method of the invention;

FIG. 2 is a block schematic diagram of an apparatus of the invention; and

FIG. 3 is a class diagram.

DETAILED DESCRIPTION

The present invention is based on the service locator pattern described in http://java.sun.com/blueprints/corej2eepatterns/Patterns/ServiceLocator.html, which is incorporated herein by reference. However, the invention extends the known service locator so that it can abstract all WS (web service) and JNDI usage and so provide more comprehensive object location than is possible using only JNDI. The service locator of the invention is still able to hide complexities of initial object creation, web service binding, EJB home land remote lookup from users, and so is user-friendly. Multiple clients can reuse the service locator to reduce code complexity, provide a single point of control and improve performance by providing a best available protocol call optimization.

The service locator uses the factory design pattern to provide a single point of control for the different objects to find. As is known the “factory design pattern” is an object-oriented design pattern that can create objects without specifying the exact class of object that is to be created. More information about the factory design pattern is available at, for example, http://en.wikipedia.org/wiki/factory_Method_Pattern.

In an advantageous implementation, the service locator uses a simple algorithm that will attempt to return the best possible object available by carrying out the following three steps:

1. Try to locate a local object;2. if 1 failed, try to locate a remote object;3. if 2 failed try to locate a remote web service object

In a particularly advantageous embodiment an object located in response to a user request is cached. In this embodiment the algorithm may further include, before carrying out step 1 of trying to locate a local object, determining whether the requested object is already available (because it has been cached following an earlier search) and, if a cached copy is available, determining whether the cached copy is still up-to-date. If an up-to-date cached copy of the requested object is available, the cached copy would be returned and there would be no need to carry out steps 1, 2 or 3. This is described further below.

In another advantageous implementation, the protocol call optimization of the service locator is configurable, and if no optimization is selected then web services are the default—that is, if no optimization is selected the service locator would not try steps 1 and 2 above, but would go straight to step 3.

FIG. 1 is a schematic block flow diagram showing an embodiment of the invention.

Initially, at step 1, a Service Locator Factory (SLF) receives a request for an object from a client 1. The object is identified in the request by a web service address such as a URL. In FIG. 1 the request is shown as identifying the type of web service as well as identifying the web service itself, but the invention is not limited to any specific format for the request.

In FIG. 1 the request is also shown as indicating than an optimisation protocol of the invention should be used, by the field “boolean optimisation” in the request. If the flag indicates that no optimization is selected the service locator would not try steps 1 and 2 above, but would go straight to step 3 of attempting to locate a remote web service object—this allows, for example, a system administrator to disable the use of the optimisation protocol of the invention if desired.

In an advantageous embodiment, the SLF 2 keeps a cached copy of objects that it retrieves as to optimize calls from other clients. If the SLF does maintain a cache, upon receipt of a request for an object it checks (step 2) whether the requested object is present in the cache and whether any parameters of the requested object have changed (for example the url). If the requested object is present in the cache and the parameters are identical this indicates that the cached copy of the object is still up to date, and the cached copy of the object may be returned to the client 1 at step 15. However, if the requested object is present in the cache but the parameters have changed since the object was cached, this indicates that the cached copy of the object is no longer up to date and so is not suitable to be returned to the client 1.

If the SLF has never previously retrieved the requested object, the result of step 2 will of course be that the requested object is not present in the cache.

If the SLF does not maintain a cache, step 2 is omitted.

If step 2 does not locate an up-to-date copy of the object, or is not carried out, the SLF 2 then proceeds to create a service locator SL 3, at step 3, and sends a request for the service requested by the client to the SL at step 4. Creation of the SL 3 is generally as set out in http://java.sun.com/blueprints/corej2eepatterns/Patterns/ServiceLocator.html.

The method then checks at step 5 whether the SL is configured to use protocol call optimization. If the SL is configured to use this optimisation (if “optimise is true”), the method attempts to locate the requested objected using JNDI look-up initially and then searching for a remote web service object if the JNDI look-up does not locate the object.

In the embodiment of FIG. 1, if step 5 determines that the SL is configured to use protocol call optimization the method applies the three-step algorithm described above. That is, the method attempts to find the local reference object through the SL by initially performing a lookup for local object using JNDI. To do this an initial context 4 is created at step 6 or alternatively, depending on the language and the framework, a link to the initial context may be created (for example with EJB 3.0 a link is injected). As a further alternative, the initial context may be retrieved by performing a new InitialContext routine with the EJB container. Next, a local object reference is created at step 7, and a JNDI lookup for a local object is carried out at step 8. If the object is retrieved, it is returned to the SLF 2 at step 14.

In a modification, it would be possible to select not to use JNDI in the local object look-up, and instead bind to a local object available within the Java Virtual Machine thereby bypassing JNDI.

If the lookup for a local object at step 8 is unsuccessful, the SL then performs a lookup for a remote object using JNDI. To do this a remote object reference is created at step 9, and a JNDI lookup for a remote object is carried out at step 10. If the object is retrieved, it is returned to the SLF 2 at step 14.

If the JNDI lookup for a remote object at step 10 is unsuccessful, the SL then determines at step 11 to perform a lookup for a remote web service object. A link to the a web service client factory (for example to the javax.xml.ws.Service http://java.sun.com/javaee/5/docs/api/javax/xml/ws/Service.html) is created at step 12 and a port is obtained at step 13. If the object is retrieved, it is returned to the SLF 2 at step 14.

At step 15 the retrieved object is returned to the client 1.

If the SLF maintains a cache, any object returned to the SLF at step 14 may be added to the cache.

One important advantage of using the Service Locator 3 is implicit optimization of calls between applications collocated in the same J2EE (Java 2, Enterprise Edition) container. Using the service locator, a request is only resolved as a web service search if a local object or a remote (RMI—Remote Method Invocation) object is not available. Measurements on JBoss show that invoking a simple web service object location request with only one argument is on average 70 times slower than a local call, and 12 times slower than an RMI call. Latency is also improved, as the CPU used when optimizing is reduced when compared to an equivalent WS call.

A further advantage is that the Service locator makes the optimization protocol transparent to the user. The user simply requests an object and (hopefully) receives the requested object, and is unaware of the method (ie, JNDI look-up or remote web service object look-up) used to retrieve the object.

The method of FIG. 1 may be performed in any suitably programmed network node, and in general is carried out at an Application Server (AS) that can implement the SLF. No new hardware is required to implement the invention.

FIG. 2 is a block schematic diagram showing a network 8 containing a node 6, for example an application server 6, that can implement the method of the invention. A node 1 that requires an object sends a request to the application server 6. The node 1 is shown as external to the network 8, but the node 1 could alternatively be within the network 8.

When the application server 6 receives the request from the node 1, and executing the request would involve an external service invocation, the application server 6 will attempt to locate that object in accordance with a method of the invention as described above. In an advantageous embodiment, the application server 6 will attempt to locate the object using the optimisation protocol described above in which the application server 6 initially looks for a local object. If a local object cannot be found, the application server 6 then looks for a remote object, that is an object located at another node of the network 8 (as schematically indicated in FIG. 2 by node 7). If a remote object cannot be found, the application server 6 then attempts to locate a remote web service object (as schematically indicated in FIG. 2 by the application server 6 launching an enquiry to the Internet 9).

In FIG. 2, the node 1 is a client of the application server 6, and the application server 6 is a client of the node 7 or the Internet 9. In the embodiment of FIG. 2, the application server 6 implements the SLF 2, SL 3, Initial Context 4 and web service client factory 5 of FIG. 1.

FIG. 3 is a class diagram illustrating the invention. In FIG. 3:

• • AServicelnterface is the base interface used by local or remote or WS implementation class.
  • AServiceBean is the Session Bean implementation which can be a remote or local EJB.
  • AServiceWS is the WebService implementation.
  • AServiceServiceLocator contains the necessary code to perform the specific code for this service (AService). It is mostly a holder of constant values such as the service name and JNDI name.
  • ServiceLocator is an abstract class that contains the logic common to all sub-interfaces.
  • ServiceLocatorFactory is an abstract factory (from the abstract factory design pattern).
  • WSType (not shown in diagram) is an enum list of all possible ServiceLocator implementation (this simplifies the look up and reduces casting errors if using String for example).

Note that it is possible to have AServiceBean and AServiceWS as the same class using EJB 3.0 annotations.

For the class diagram of FIG. 3, invoking:

ServiceLocatorFactory->getService(WSType type, boolean optimize, String wsUrl) throws MtvConfigurationException (MtvConfiguration is a Java Exception, and may be replaced by java.lang.Exception or simply Exception. The class Exception is a subclass of the Throwable class in the Java language, and is conventionally used to indicate that an exceptional situation has occurred.) This will return the optimized implementation of the service AService.

In addition the ServiceLocatorFactory may keep a cached copy of the latest obtained service so as to optimize calls from other clients.

On an attempt to retrieve the AService object, one of the methods described above is followed.

The present invention may be employed generally to locate an object across a network, and is not limited to any one particular use. It may be used in a mobile system, or in a fixed system.

Claims

1. A method of locating an object across a network, the method comprising: locating the copy of the object that is the best copy available through either JNDI object location and webservice location.

2. A method as claimed in claim 1 and comprising, in response to a request for an object: carrying out a Java Naming and Directory Interface (JNDI) search for the object;if the JNDI search locates a copy of the object, retrieving the copy located by the JNDI search; orif the JNDI search does not locate a copy of the object, attempting to locate a remote web service object.

3. A method as claimed in claim 2 and comprising, if a remote web service object is located, retrieving the copy of the remote web service object located.

4. A method as claimed in claim 2 wherein carrying out the JNDI search comprises: attempting to locate a local copy of the object;if a local copy of the object is located, retrieving the local copy of the object; orif a local copy of the object is not located, attempting (step 10) to locate a remote copy of the object.

5. A method as claimed in claim 4 and comprising, if the search locates a remote copy of the object, retrieving the remote copy of the object.

6. A method as claimed in claim 1 and comprising attempting to locate a requested object in a cache before attempting to locate the object through either JNDI object location or webservice location.

7. A method as claimed in claim 1 and comprising caching a retrieved object.

8. A network node for locating an object across a network, the node being adapted to locate the copy of object that is the best copy available through either JNDI object location and webservice location.

9. A network node as claimed in claim 8 and adapted to: carry out a Java Naming and Directory Interface (JNDI) search for the object;if the JNDI search locates a copy of the object, retrieve the copy located by the JNDI search; orif the JNDI search does not locate a copy of the object, attempting to locate a remote web service object.

10. A network node as claimed in claim 9 and adapted to carry out the JNDI search by: attempting to locate a local copy of the object;if a local copy of the object is located, retrieving the local copy of the object; orif a local copy of the object is not located, attempting to locate a remote copy of the object.

11. A computer-readable medium containing instructions that, when executed on a processor, perform a method of locating an object across a network, the method comprising: locating the copy of object that is the best copy available through either JNDI object location and webservice location.

12. A computer-readable medium as claimed in claim 11, the method comprising, in response to a request for an object: carrying out a Java Naming and Directory Interface (JNDI) search for the object;if the JNDI search locates a copy of the object, retrieving the copy located by the JNDI search; orif the JNDI search does not locate a copy of the object, attempting to locate a remote web service object.