|
To answer that question and others, Geoffrey Greene, Senior
Software Engineer for ColdFusion MX for J2EE, sat down to
talk about the architecture underneath ColdFusion MX for
IBM WebSphere J2EE Application Server and discuss how it
will impact your web application development.
ColdFusion MX Server already ships with an integrated
J2EE server. How is the ColdFusion MX for J2EE release different?
First a little background. ColdFusion MX Server Professional
and Enterprise Editions—I refer to these as the standalone
editions—are built in Java and include an embedded
J2EE server, but for the most part the underlying Java functionality
is invisible to the developer. By contrast, ColdFusion MX
for IBM WebSphere [and the other ColdFusion MX for J2EE
products] gives you access not only to ColdFusion functionality
but lets you combine it with all of the features of your
Java app server. For instance, you can integrate servlets
and JSP pages into your ColdFusion pages, leverage the underlying
J2EE application server functionality (such as clusterable
sessions and load balancing), share data between JSPs and
servlets through a request object, and access EJBs and other
Java resources with the cfobject tag.
On which servers does ColdFusion MX for J2EE run?
Right now, we support IBM WebSphere versions 4.03 [Advanced
Edition and Advanced Edition Single Server], Sun ONE Web
Server, and Macromedia JRun 4.0. Later this year we plan
to add support for IBM WebSphere 5, BEA WebLogic 7.0, and
Sun ONE Application Server 7.
How is ColdFusion MX for J2EE installed on a J2EE
app server?
ColdFusion MX for J2EE is packaged as a [Web Application
Archive] file. The ColdFusion MX installer unpacks the WAR
file and then calls the application server's deployment
facilities. The deploy directory is fixed and determined
by the underlying application server. For IBM WebSphere,
it's located at /AppServer/installedApps/cfusion.ear/cfusion.war.
The runtime files for ColdFusion MX for J2EE are nested
in a directory structure underneath the cfusion.war directory.
The /web-inf/lib directory holds all of the JAR files for
the ColdFusion runtime, and /cfusion holds all specific
ColdFusion files. The cfusion.war directory is also where
ColdFusion places all of the Java class files it creates
when it compiles your application's ColdFusion templates
into Java bytecode.
Do you have to rewrite CFML pages from existing
ColdFusion apps to run on ColdFusion MX for IBM WebSphere?
For the most part, no. ColdFusion MX for J2EE supports the
same language elements as ColdFusion MX Server. If you move
applications to ColdFusion MX for J2EE, there may be minor
changes, such as context root issues and Verity tweaks,
but the application logic should function just as it does
with the standalone version.
How do developers integrate existing Java assets/objects
on the app server with ColdFusion applications?
There are a few ways to integrate with Java. First, because
ColdFusion uses the servlet API under the covers, you can
easily share data with—or forward control to—Java
servlets and JSPs. Second, you can import Java tag libraries
using the cfimport tag. You just assign them a prefix and
then call them just like Custom Tags in ColdFusion. Lastly,
you can work directory with Java objects using the cfobject
tag.
How does it handle sessions?
Session storage is just the same as standalone ColdFusion
MX server. The only caveat is clusterable sessions; they
are not serializable. In order to cluster in this case,
you must put it into a portable format and make it serializable
with a keyword.
What is the performance like on these J2EE versions?
Performance really depends on your application, but some
customers have seen performance improvements because ColdFusion
MX for IBM WebSphere has been optimized to take advantage
of the high-performance features in WebSphere, including
their database connection pooling and their latest JVM.
How are CFML page compiled in this version? How
are page requests handled at runtime?
Page compilation is the same in all versions of ColdFusion
MX. The first time a page is requested, it is compiled into
Java bytecode and cached to disk. Once a page has been compiled,
subsequent requests are passed directly to the compiled
class file for processing. What's different in the ColdFusion
MX for J2EE products is that the execution of that bytecode
is managed by the application server where you installed
ColdFusion MX, such as IBM WebSphere.
What kinds of data sources are available inside
ColdFusion?
There are two kinds of data sources in ColdFusion MX for
J2EE. As with the standalone version, you can create and
modify data sources through the ColdFusion administrator.
These use a built-in pool manager that is part of ColdFusion
MX for J2EE. You can also use data sources that were defined
in the underlying app server. These can be called from ColdFusion
applications just like any other data source, but they appear
in a read-only format in the ColdFusion administrator and
can only be modified using the underlying application server's
tools.
|
