Grid and Your SOA Strategy

Grid and Your SOA Strategy by Jim Byrd

Virtualizing grid infrastructure components to provide management and analysis functionality.

Companies today face increasing IT infrastructure demands, often without concomitant increases in budget or resources. At the same time, critical client applications are integral to continuous, high-quality service. Increasing application performance and resiliency can help bolster a company’s position vis-à-vis its competitors, and keep pace with increasing or volatile service demands.

A key concept gaining traction with IT managers to affect these goals is a Service-Oriented Architecture (SOA) strategy for enterprise systems. SOA is a form of distributed computing that is both application-centric and demand-driven, focused on facilitating a key process. Service-Oriented Architecture SOA achieves this by virtualizing application components and then linking and leveraging these heterogeneous resources across the enterprise to provide optimal client service. By nature, Service-Oriented Architecture SOAs are loosely-coupled, software modules independent of implementation and infrastructure details. This allows developers the freedom to write and edit services without affecting the underlying implementation. The ease of manipulating and modifying services, while leveraging an organization’s existing resources (including infrastructure, IT staff, languages, platforms and databases), makes Service-Oriented Architecture SOA a viable strategy for many companies that already have sunk high costs in existing IT investments.

Grid Computing and Service-Oriented Architecture SOA

Grid computing is in many ways a natural complement to a SOA implementation, since many of the core requirements for a Service-Oriented Architecture SOA are naturally matched by the strengths of grid computing. Clearly, a successful Service-Oriented Architecture SOA implementation requires an accompanying service execution platform to serve as a “fabric” for virtualizing components and to provide management and analysis functionality. Grid-computing infrastructure enables organizations to quickly realize the benefits of a Service-Oriented Architecture SOA without having to undergo costly application rewrites or re-architecture. Besides allowing for global management and execution, a good service platform will also have monitor and alert functions, service level agreement (SLA) tracking and accounting capabilities, easy-to-use interfaces and message exchange protocols and dynamic provisioning. Out of the box solutions are readily available, simplifying implementation and facilitating coordination and interoperability. By using a grid-computing service platform with the above characteristics and capabilities, companies immediately observe the following benefits and help accelerate transition towards SOA:

- Application Performance, with up to a 50-times increase in application response time and throughput

- Resilience and Reliability, as failure rates can drop by 90 percent or more with guaranteed task execution and built-in contingency mechanisms to ensure against failure or system error

- Flexibility and API independence, with vendor-neutral support for a wide variety of clients including leading standards Java, .NET, SOAP, C++ and binary executables. Grid-based environments can be incremental and reversible, able to accommodate newer interfaces based on WSDL, SOAP, XML and BEPL as well as legacy applications

- Service-Oriented Control, allowing for global management and administrative control of operational parameters such as policy-driven service, resource assignment and workload distribution rules

- Dynamic Provisioning, optimally utilizing heterogeneous resources across the enterprise by dynamically and adaptively balancing loads and provisioning tasks within milliseconds

- Rapid Development and Deployment, simplifying and streamlining the production process through standards-based infrastructure that manages QoS and enforces service access and execution policies

- Usage-Based Accounting, with centralized administration tools that allow managers to establish variable cost accounting systems

- TCO Reduction, improving the bottom line through creating a more cost-efficient, productive, and reliable IT system, without additional investments in infrastructure or human resources

At the service level, grid computing can provide additional value-add through parallel execution, high concurrency of task processing, redundant task submission, and distributed caching

Benefits of Grid-Enabled SOA Service-Oriented Architecture

The benefits of a SOA strategy are almost immediately evident, especially when compared to the model most organizations use today—individual silos dedicated to each client application. This silo approach houses and deploys applications on dedicated IT systems, limiting capacity and availability, and often introducing significant latency and execution failure during periods of high compute demand. Virtualization of such systems through a Service-Oriented Architecture SOA strategy allows an organization to significantly reduce risk and improve performance and flexibility by abstracting, distributing and managing distributed resource layers. Additionally, because SOA is demand-driven, resources are used on an “as needed, when needed” basis, and tasks are appropriately assigned based on specified parameters. Virtualization allows a Service-Oriented Architecture SOA to meet fluctuating demand, increasing performance and reliability, while QoS parameters such as security, scalability and support are built-in.

Summary

Grid computing is highly complementary to the Service-Oriented Architecture SOA model. As a design concept, SOA can provide substantial gains in business performance while requiring no additional investments in infrastructure or human resources. A comprehensive grid service platform can act as an appropriate way to achieve the goals of a Service-Oriented Architecture SOA strategy by 1) virtualizing the service execution environment; 2) globally and automatically managing service execution; and 3) virtualizing the underlying system and data-level resources. These benefits confer exponential increases in scale, price and performance gains, allowing companies to more effectively meet the challenges of a changing global marketplace.