Architecture

Appistry EAF is a grid-based application platform that dramatically simplifies the development and deployment of scalable applications in Java, .NET or C++. Unlike traditional application servers, Appistry EAF is built from the ground up for scalability. And unlike traditional grid computing, it's extremely easy to fabric-enable your applications.

Service Access

Your application fabric--no matter how large it becomes--looks like a single virtualized system to the clients and systems utilizing its applications. Appistry's Fabric API gives .NET, Java and C/C++ client applications native access to fabric applications and data. Through Appistry-provided integrations, customers may also access fabric applications via Remoting and Web Services.

Appistry EAF provides software-based load balancing by distributing work across the underlying grid. Load balancing occurs in a fully distributed manner--there is no "master node" or single point of failure. Load balancing is also adaptive, taking into account the capacity of individual nodes in a heterogeneous grid.

Distributed Container (Compute Grid)

Appistry EAF scalably and reliably distributes application components across an underlying compute grid. Developers can easily scale-out existing components and binaries with no changes to code, or utilize Appistry EAF APIs to build distributed caching or fine-grained reliability into their mission critical applications.

Developers can take advantage of a simple, process-based abstraction that eliminates the complexity of creating robust, highly-distributed applications and allows them to focus on their business problems. Using existing tools, such as Eclipse or Visual Studio, developers build application components (POJOs, PONOs, etc.) and define orchestrations of these components in XML. The Appistry EAF runtime transparently provides reliability and scalability to applications running in the fabric.

A variety of workload management policies ensure that applications--whether transactional or computational in nature--receive the resources needed for maximal performance.

Distributed State (Data Grid)

Fabric Accessible Memory (FAM) is a distributed, in-memory data grid providing reliable data caching for application state information. Architects and developers utilize FAM to:

bolster application performance by reliably caching application state in-memory
reduce database bottlenecks by utilizing RAM for working storage
simplify data sharing across disparate applications and application components

FAM features such as events and queues facilitate distributed message passing and enable the implementation of a variety of architectural patterns including blackboard, master-worker, scatter gather, map reduce and others.

Because FAM is built on top of the application fabric, it inherits its characteristics such as reliability, availability, scalability and flexibility.

Policy Definition & Enforcement

Workload management policies enable an application fabric to flexibly support a variety of CPU- and data-intensive application workloads.

These policies provide for the timely completion of fabric requests by allowing workloads to demand additional compute capacity as needed. With policies in place, Appistry EAF is able to optimally allocate resources to demanding applications, and better manage application performance based on workloads and events.

In addition, the fabric's affinity feature enables faster, more powerful applications by allowing data to be partitioned and distributed to individual computers within an application fabric. Requests processed by the fabric--e.g. transactions, jobs, events or service invocations--are then dynamically routed to the machine containing the appropriate data.

Management & Monitoring

Appistry's web-based Fabric Monitor provides a simple interface for monitoring fabric performance and capacity. In addition, Appistry EAF publishes a variety of key performance indicators via a Web service, which is readily consumed by enterprise management platforms.

Deployment & Provisioning

Traditional approaches to application scalability stress an organization's operational infrastructure and staff as well as its architects and developers. Appistry EAF offers a comprehensive, holistic approach to application scalability that addresses operational and management considerations by automating deployment and provisioning of new fabric workers.

Because the physical infrastructure underlying an application fabric is virtualized, system administration remains simple, independent of fabric size. To update any software component application, administrators simply deploy the component (e.g. the application, fabric software, etc.) to the fabric as a whole which in turn ensures its proper deployment on individual servers.

Additional features include:

Discovery: Appistry EAF automatically discovers bare metal hardware added to an application fabric as soon as a network connection is established.
Bare-metal provisioning: Appistry EAF ensures that appropriate versions of the operating system and fabric software are deployed on the new resource
Assimilation: Newly provisioned resources are automatically introduced into the application fabric and begin serving application requests
Rolling updates: Administrators can deploy application updates to a production fabric without taking the application off-line by rolling the update across the fabric. Appistry EAF coordinates the update, systematically taking machines off-line, applying the update, and reintroducing them into the application fabric until all machines are updated.