News Stay informed about the latest enterprise technology news and product updates.

A Look at the Apache Tuscany Project

The Apache Tuscany project provides multiple language implementations of Service Component Architecture (SCA) In this Q&A, hear how Tuscany works and what its goals are.

by Colleen Frye

The Apache Tuscany project provides multiple language implementations of the Service Component Architecture (SCA) specifications and related technologies. An upcoming book, Tuscany SCA in Action, is a guide for developing technology-agnostic, service-oriented composite applications using Apache Tuscany's implementation of SCA. It was co-written by Mark Combellack, Raymond Feng, Simon Laws, Haleh Mahbod and Simon Nash. In this interview, two of the authors, Mahbod and Feng, speak about Tuscany and SCA. Mahbod is an Apache Tuscany PMC member and committer. She has led development organizations that delivered SOA-based integration platforms and solutions in open source, in Fortune 500 and start-ups. Feng is an Apache Tuscany PMC member and committer and a senior engineer at IBM. Tell us about Tuscany, an implementation of Service Component Architecture (SCA) and an open source Apache project.

Haleh Mahbod: You need a common way to describe an assembly of distributed services regardless of the technologies used, and to do so without polluting the business logic. That's what SCA provides, and what Tuscany implements: a way to follow a SOA paradigm and use SCA as a programming model to assemble these services. To take a step back, really Tuscany is composed of two major pieces: The infrastructure side of it, which frees the application from being polluted from extra code; and implementing a standard language for SOA, which gives you flexibility and portability. The architecture is very modularized. You can take pieces of Tuscany and embed it in your own software.

Tuscany 1.x is a full implementation of SCA 1.0 developed by [the Open SOA Collaborative, which developed version 1.0 of the SCA spec, before turning it over to OASIS]. The OSOA version 1.0 is pretty mature now, and it has many users. SCA 1.1 is in the process of being standardized at OASIS. The Tuscany 2.x version is implementing that, and we'll be helping the standards team with compliance testing for SCA 1.1 to make sure the 2.x version of Tuscany is compliant with the SCA 1.1 version.

SSOA: What was the impetus for Tuscany?

Mahbod: We at IBM started developing SCA. At the time WebSphere, the process server and the ESB were based on SCA. And then we collaborated with BEA and [later] the rest of the members of OSOA and later OASIS to create a standard. For Tuscany itself, the goal was to introduce the technology to the market globally. Also we behaved as a testbed for the spec. We did an early implementation for the spec, and we provided feedback [from the developers] and user feedback from people using Tuscany, strengthening the implementation. The other thing Tuscany did, it was an innovation test for spearheading new projects—we have an Android implementation, and we've experimented with cloud computing on Google Apps, for example.

Raymond Feng: I think innovation goes beyond adding extensions to what's not been specified in the SCA spec. Tuscany expands SCA in that direction, with new extensions to cover popular programs/frameworks. Tuscany also identifies new opportunities for SCA, not just purely as a programming model but also how it fits to new opportunities, such as integration between SCA and the OSGi enterprise pack.

SSOA: Is OSGi, a component integration platform and dynamic module system for Java, complementary to SCA?

Feng: When people talk about OSGi they're possibly talking about different things. One is as a Java module assistant to make a complex platform more manageable. It also defines a service programming model that allows you to use Java. Tuscany covers OSGi from both perspectives. It has a runtime implementation of the SCA spec, so it enables Tuscany to host OSGi bundles. In Tuscany, you make OGSi bundles viable as SCA components.

Mahbod: SCA extends OSGi with its high-level, coarse-grained composition. It enables you to incorporate components that are not Java based.

SSOA: Microsoft hasn't been part of the SCA effort. How has Tuscany addressed this?

Feng: From the spec side, even though Microsoft is not involved, they have made some comments in public review. At this point I don't see any implementation of SCA on a Microsoft platform, but it doesn't exclude that. There has been some discussion, but there is no contribution coming from the [Tuscany] community yet.

SSOA: What will readers take away from your book, Tuscany SCA in Action?

Mahbod: The book is a tool for learning SCA for the first time, with detailed practical examples for creating composite applications from services. The main example in the book is a travel booking application. You start implementing small components and grow to [implementing larger components]. All the source code is available in the Tuscany project. It's intended for developers who care about creating reusable services and assembling them. We focus on the freedom of choice of technology for developing components, choosing protocols [and] the hosting environment. We show how easy it is to use SCA. Part four talks to people who want to extend Tuscany and incorporate any technology we might not be providing out of the box. We provide many out of box.

Your takeaway is you'll understand SCA and how useful it is to business. You'll gain hands-on experience developing a travel booking application, and from that you can jump-start your own business application. We have best practices outlined—how to work with Apache Tuscany, how to get help, how to participate.

Dig Deeper on Topics Archive

Start the conversation

Send me notifications when other members comment.

Please create a username to comment.