Oracle reveals Project X for application integration
Oracle recently announced Project X, a new application integration strategy for its disparate portfolio of applications.
The project, announced at Oracle's Collaborate user forum in Las Vegas, could reduce the complexity of integration, but could also detract from its forthcoming Fusion applications because it provides a way of streamlining interactions and process flows across heterogeneous applications.
At the heart of the new architecture is a translation layer that sits between the applications and Fusion Middleware and is accessed via said middleware, and an emerging set of common definitions and objects that applications can conform to. The approach caters for the different ways applications define objects such as a sales order.
With no standard definition, each application needs to know how the other applications it integrates with defines an order so it can expose it in a meaningful way, explained Paco Aubrejuan, vice-president of application strategy at Oracle. This means that even in a service-oriented environment, if an application integrated with multiple applications it would have to expose multiple versions of the same service.
Paco Aubrejuan, Oracle vice-president of application strategy
The Oracle approach is to have applications conform to a common set of object definitions. An application could expose an order based around a common definition, making it available to others through an intermediary framework. Consuming applications should be able to pick up the common definition and transform it into the format they need.
Using this approach, Aubrejuan said, some of the cost and complexity of application integration can be reduced because each application has only to expose one standard version of a service that conforms to common definitions, rather than having to expose a different version for each application.
One of the advantages of this service-based framework-style is that it allows disparate applications to work together without demanding changes to the application architecture, making it attractive to Oracle with its portfolio of acquired and in-house developed applications. It should also make upgrades easier, Aubrejuan says.
"Instead of changing the data models, which is not feasible, we are creating a single logical representation of an order that all systems adhere to when taking an order," he said. "It reduces the complexity of integrations and the cost of maintaining integrations."
Aubrejuan said that Oracle was using this approach internally to enable integration across its disparate applications, and was making it available to customers. Business application consolidator Infor has also adopted a framework type of approach to deal with its disparate applications.
Fujitsu and Sun deliver joint Sparc Enterprise Server line
Timothy Prickett Morgan
Sun Microsystems, Fujitsu and its Fujitsu-Siemens joint venture last month delivered what is being pitched as a jointly developed Sparc server product line called the Sparc Enterprise line.
How much joint development was done on the Sparc Enterprise servers is debatable. The entry products are essentially rebranded Sun T1000 and T2000 servers, which use the Niagara Sparc T1 multicore processor. In this case, joint development seems to mean spray painting them in the company colours.
At the other end of the Sparc Enterprise server line, Fujitsu seems to have done all of the work, designing the dual-core Sparc64 VI Olympus chips and the Jupiter server frames that house them. What the Sparc Enterprise servers are, in truth, is a blending of two existing Solaris-compatible product lines from two companies that sometimes partner and sometimes compete.
The Sparc Enterprise line is the result of a partnership agreement that Sun and Fujitsu signed in June 2004,when Sun had scrapped its own UltraSparc V Millennium chip and was looking for a stop-gap to carry it through to the Rock multicore and multithreaded processor and the related Supernova servers, which are coming to market in 2008 - on time, as far as anyone can tell.
Fujitsu had originally planned to launch its own PrimePower machines using the Sparc64 VI chips in mid-2004 or so, and for whatever reason, the product was delayed just about the time that Sun came looking for a partnership.
What is clear now is that the Advanced Product Line partnership, as this was code-named, was a means to give both Sun and Fujitsu a story to tell while they were waiting for their chips and servers to move from development to products. After the Sun-Fujitsu partnership was launched nearly three years ago, the companies said they expected to have the high-end servers to market by the middle of 2006. They are therefore nearly a year late coming to market.
The new machines in the Sparc Enterprise line include the midrange line, the M-Series, and a high-end line, which are based on the long-awaited Fujitsu Sparc64 VI chip. These chips run at 2,15 GHz, 2,28 GHz,and 2,4 GHz, and they have two threads per core, for a total of four cores per CPU socket. That is as good as IBM and Intel can deliver with their high-end chips, and better than Sun can do with its UltraSparc-IV+chips.