Operational Excellence for Distributed, Interconnected Applications

WHITE PAPER
SOA OPERATIONS EXCELLENCE
WITH PROGRESS ACTIONALTABLE OF CONTENTS
> 1.0 Service-Oriented Architecture: Benefits and IT Operations Challenges 1® ®> 2.0 Progress Actional for SOA Operations 4
2.1 Automatic Service Discovery 5
2.2 Non-Intrusive Runtime Monitoring 6
2.3 Policy Management 7
2.4 Policy-Based Alerting 8
2.5 Problem Resolution 8
> 3.0 Advancing SOA with Actional 9
Copyright ©2007. Progress Software Corporation. All rights reserved.1.0 SERVICE-ORIENTED ARCHITECTURE: BENEFITS AND IT OPERATIONS CHALLENGES
Service-oriented architecture (SOA) is the systems and software approach of choice for1the majority of medium and large enterprises. According to the principles of SOA, ITprofessionals design, implement, and deploy information systems from components thatperform discrete business functions. These components, called "services," can bedistributed across geographic and organizational boundaries, can be independentlyscaled, and can be reconfigured into new business processes as needed. As well as thebenefit to business and IT of such flexibility, SOA easily allows services to be accessibleover the Web, allowing the creation of new on-line businesses and product offerings.
At the same time, IT organizations in every industry sector must meet exceedingly highstandards for systems reliability, cost-effective operations, and customer service.Competitive pressures and the needs of both internal and external users allow no roomfor mistakes in IT operations and applications management. Many are looking to2approaches, such as ITIL, specifying best practices for IT management. ITIL defines, forexample, the need for the operations and applications groups to provide service-levelmanagement, requiring them to meet specific service-level targets (using the broadermeaning of "service"), ranging from response time targets for interactive systems to thetime it takes to diagnose and correct a problem. Figure 1 shows some sample keyperformance indicators (KPIs) suggested for a subset of the ITIL management areas.
3Figure 1: Sample ITIL Key Performance Indicators
Service-level Management
> Percentage reduction in SLA (Service-level Agreement) targets missed.> Percentage reduction in the service delivery costs.
Availability Management
> Improvement in the MTBF (mean time between failures).> Percentage reduction in the unavailability of services and components.
Incident Management
> Percentage increase in the incidents fixed before users notice.> Percentage increase in the incidents resolved by first line operatives.
1Aberdeen Group, Management and Governance: Planning for an Optimized SOA Application Lifecycle, March 2007.http://www.actional.com/resources/whitepapers/.2Office of Government Commerce, http://www.itil.co.uk/.3Office of Government Commerce, Best Practice for Planning to Implement Service Management, The Stationery Office, Norwich, UK, 2002.
Copyright ©2007. Progress Software Corporation. All rights reserved. 1The distributed, heterogeneous, and "loosely-coupled" nature of SOA as well as newtechnologies and protocols such as the ESB (enterprise service bus) and SOAP createnew challenges requiring new approaches to the monitoring and problem diagnosisrequired for IT operations best practices. Traditional systems management tools, whichare focused on performance and availability at the hardware and operating system level,lack the capabilities to manage SOA systems and applications. For example, considerthe following SOA operations management requirements, which traditional systemsmanagement tools, even with application support and "SOA features," are not able tofulfill:
> A hallmark of SOA systems is that services are shared-reusing standard businessfunctionality. However, the performance of the shared service may differ according tothe context in which it is used. For example, the average response time of a "creditcheck service" might be 45 msec. when used by an external Web service accessibleto other financial institutions, but 10 msec. for an internal requester that calls thisservice twice as often. In traditional tools, an average response time SLA (service-level agreement) of 30 msec. for the "credit check service" would never indicate thata consumer of the service was having a performance problem. Consequently, theremust be the ability to define and mo... [download for more]