FRAMINGHAM, 13 JANUARY 2009 - There's a growing desire among IT professionals to manage their diverse technologies according to the services they deliver, to have a predictive view of those services and a cross-silo view of the components that comprise them. This is driven by the need to prevent problems before they affect the business, quickly pinpoint root cause, put an end to the finger-pointing between the IT teams responsible for each silo, and speed mean time to repair.
Best-practice management tools can provide a "service model," a software representation of all the IT elements that are required to deliver a service. Besides visualization, the service model should have monitoring hooks into actual elements, an understanding of the relationships among those elements, policy-based thresholding for proactive warnings/alarms, event correlation, root cause analysis and alarm (symptom) suppression.
How you define an IT service depends on personal preference. An IT service may be CRM, an online shopping cart, express mail tracking, insurance benefits selection or e-mail -- essentially what the users see or experience. In all cases, the service model is comprised of a particular cross-silo collection of IT components: network devices, physical and virtual systems, databases and applications. Each component's health is monitored and its status is reflected in the service model to show its current or projected impact on the service.
IT organizations are eyeing IT service modeling because it enables them to manage technologies in the context of business impact and to prioritize corrective actions according to business priority. Because IT service modeling is a grouping of IT components, it requires the sharing of data across siloed management tools, teams and processes.
Integrated network and application management is essential. To show all IT service dependencies, IT modeling integrates infrastructure management for all silos (networks, systems and databases) with application management. In real time for each IT service the model shows the current status, trends and root cause based on: 1) real-time measurement of user experience; 2) a comparison of the time each business transaction spends in each silo as it traverses the infrastructure; and 3) health analysis of each infrastructure component and of the application itself.
User experience measurement indicates service-level agreement (SLA) compliance. End-to-end monitoring shows where (that is, in which silo) a business transaction's performance degrades or outright fails. Health analysis of each silo reveals how the underlying infrastructure may be negatively impacting the business transactions. Health analysis of the application reveals, for example, how the J2EE or .Net software environment itself may be negatively impacting the business transactions.
Infrastructure monitoring may show uptime and performance within accepted tolerances (operating level agreements), yet some applications may be violating SLAs due to poor software configuration or constrained database server resources. Hardware server utilization may be within prescribed limits, yet certain applications may be hogging server resources enough to throw others above their prescribed response times. Infrastructure may be blamed for poor performance, yet the application software itself may be poorly written.
Sign up for Computerworld eNewsletters.