Application fabric is a term that describes software that distributes tasks to dissimilar, potentially budget-priced commodity servers while delivering the high availability that Oracle Corp. touts with RAC (Real Application Clusters).
Application fabric technology, which lives on the application layer, ties together the benefits of scalability, agility and low cost that are attributed to grid computing and virtualization.
The company announced Appistry Enterprise Application Fabric 3.0 here at Gartner Symposium/ITXpo. The updates new FAM (Fabric Accessible Memory) is built on top of the application fabric, in which every piece of data stored to FAM is located in memory on multiple computers.
The update also comes with a new framework that lets developers define events. Developers can now propagate events across the application fabric. Events and FAM can work together to form a "bus" that allows services or application fabric tasks to communicate within the fabric.
Also new in 3.0 is a queuing feature that allows developers to define policies to establish the sequencing model of applications. Queuing can also be exposed via APIs, enabling advanced queuing, sequencing and event processing applications to be constructed.
EAF 3.0 also comes with operational enhancements. For example, updates to the application fabric no longer require a system image to be created.
With EAF 3.0, application fabrics also are now "version aware," meaning that when new servers come on to the fabric, their application versions are automatically updated if necessary.
EAF 3.0 also now supports Novell SuSE Linux 9, in addition to the platforms already supported: Windows XP, Windows Server 2003 and Red Hat Enterprise Linux 3.
Appistry isnt the only vendor to be taking the grid concept to the application layer. DataSynapse Inc. rolled out its FabricServer software earlier this month at GridWorld. FabricServer is similar to EAF in that its designed to virtualize transactional applications within a shared grid environment, thus granting better performance, resiliency and manageability.
According to Appistry executives, where Appistrys product differs is with its application-level fault tolerance capability, which ensures that all transactions submitted to the fabric get reliably completed.
"This is very different from the increases in availability offered by traditional clustering [and in turn by DataSynapse]," said Sam Charrington, vice president of product management and marketing. "These types of solutions typically leave reliability up to the developer to implement, which can be very difficult, or expect to be run on expensive fault-tolerant hardware."
In addition, with EAF, administrators see a fabric as one thing, he said. "An application fabric is typically a number of commodity-grade machines in a data center, and the fabric software allows you to manage them as one entity," Charrington said.
"The grid solutions offered by DataSynapse [including their FabricServer] are more about provisioning applications on, and allowing companies to push work out to, existing boxes … that are underutilized."
Appistry also maintains that the new FAM capability in 3.0 is a big differentiator and is unmatched by DataSynapse products.
"It helps illustrate the different approaches. With 3.0 Appistry is offering customers an easy way to build very reliable, scalable and easily manageable applications and services, including stateful applications," he said.
EAF 3.0 will be available in a preview release in November. The product is expected to be generally available in the first quarter of 2006.