Search, View and Navigation

New Wave in IT Performance Monitoring

Netuitive Self-Learning Technology: The New Wave in IT Performance Monitoring

The inherent flaw with conventional monitoring tools and even newer analytics products is that they rely on human assumption. These tools require administrators to program logic into their performance models through the use of rules and scripts. Moreover, they require operators to set static thresholds in order to generate monitoring alerts. With all of this guesswork it is no wonder why IT organizations are primarily reactive, always fire-fighting and remain at the mercy of their business systems - never knowing where things will break next.

While this has mostly been accepted as the industry norm, we are now in the midst of several transformations that make it impossible to stay on the same path while remaining competitive. First, there is the always-on enterprise system slowdowns are no longer an inconvenience, they're a threat to business reputations, operations and the bottom-line. Second, there is the movement toward ITIL and business service management operating IT as service delivery organizations rather than standalone system silos. And third, the rapid emergence of virtualized servers, where performance management is even more complex than it was on the physical ones.

The increasing operational demands and complexity created by the always-on enterprise, cross-silo IT service management and virtualization are quickly making the rules-based tools increasingly less effective. While some improved products have emerged to address this challenge such as "event correlation," "dynamic thresholding" and "pattern matching" tools, Netuitive has made a leap forward with its self-learning performance management technology.

Rather than depending on human guesswork, Netuitive uses a statistical-based approach which automatically analyzes and correlates thousands of system metrics in real-time to learn the normal behavior patterns of a given environment, provide an end-to-end service health dashboard, isolate root-causes and forecast degradations.

Netuitive: Taking Guesswork out of the Equation

Netuitive software leverages existing monitoring agents, such as Tivoli, to collect raw numeric data at the sub-system metric level for each key performance indicator (KPI), such as CPU and memory utilization, context switching, disk and I/O activity and hundreds of other application metrics. Netuitive learns the behavior patterns of each individual KPI for a given day of week, hour of day and minute of the hour. It also learns how one KPI behaves in context of the others, which is essential for gaining a holistic picture of system and service health.

Contextual Analysis

Contextual understanding of KPI behaviors through an objective lens is fundamental for effective performance management. As an analogy, a medical doctor determines his patient's health based on a combination of vital signs (key performance indicators) such as blood pressure, heart rate, body temperature and others. Only by observing and analyzing (correlating) all of these conditions together can the doctor accurately diagnose the patient's health and even predict future illness.

Similarly, Netuitive assumes nothing about the "patient." Instead it analyzes and determines outcomes based on its own observation. Like an automated diagnostician, this technology continuously analyzes IT "vital signs," to determine the current and anticipated health of systems and the services being supported.

Netuitive self-learns the interdependencies between each "vital sign." It understands how each KPI influences the other, which can be seen in the software's "Correlation Assistant" interface (See diagram). All of these interdependencies, which are scored on a percentage basis from 0.0 to 1.0 are self-learned. None of these correlation coefficents are determined through manual means.

To determine how each KPI behaves in context to the others, Netuitive uses multivariate regression analysis and other mathematical techniques to predict outcomes of a given KPI through the observation of other related KPIs. For each KPI, Netuitive calculates its "contextual" performance value in real-time

All three methods - actual, contextual and forecasted -- are calculated simultaneously to understand overall system health. By using multiple, simultaneous methods of analyzing each individual KPI, Netuitive delivers unrivaled accuracy for IT system management.

Trusted Alarms

Trusted Alarms are one of Netuitive's most valued features and only made possible through Netuitive's automated contextual analysis approach. Delivered on both a real-time and forecasted basis, these alerts provide an easy-to-understand composite view of impending issues, and can be integrated into existing monitoring consoles and trouble ticketing systems. They are generated using an accumulated score of of real-time, contextual and forecasted deviations - taking into account the number, frequency and severity of each deviation. Each system-verified Trusted Alarm results from analyzing dozens of real-time and forecasted calculations. A Trusted Alarm can be generated for an individual system issue or an end-to-end business service. When compared to alerts generated from manual static-thresholds, Netuitive Trusted Alarms are not only proactive, but reduce false alerts by a dramatic ratio.

Health Index

System and service "health" indicators are only made possible through the use of contextual analysis. Just as a person's overall health can be determined through observation and analysis of multiple vital signs, Netuitive determines health by understanding related KPIs that make up a service delivery ecosystem - business activity, applications, servers, databases, network, etc..

For example, if a person is hyperventilating it might be because they're having a heart attack, not a breathing problem. Or maybe this is a normal response to a vigorous jog. By correlating heart rate to breathing patterns along with other indicators, and analyzed with normal behavior patterns for the given time period, the underlying problem can be accurately diagnosed. Likewise, in IT environments, CPU spikes can be caused by a failing hard drive, a badly behaving application, a sudden flood of users or a virus attack. It could also just be a harmless anomaly. But without contextual analysis there is no easy way to determine what's really happening.

Netuitive self-learns regression weights and correlation coefficients between all the customer experience (latency measurements) and infrastructure metrics automatically and in real-time. In addition, Netuitive generates a health index by also considering the frequency and severity of a given set of anomalies. Health is represented in the software through the Netuitive dashboard and service models.

Workload Index

The Workload Index is also a powerful derivative from Netuitive self-learning. These values are automatically calculated in real-time, without the need for any manual configuration, using KPIs collected from standard monitoring tools. The Netuitive Workload Index is represented by a value between 0 and 100, and factors in multiple KPIs by resource type. As an example, for an operating system -- resources consist of CPU, memory, network and disk -- where each resource is represented by multiple KPIs. In addition to OS-related workloads, Netuitive uniquely builds workload indexes for any hardware or software component such as the application, storage area network, middleware or server clusters.

Conclusion: The Value of Self-Learning Technology

While today's businesses rely on complex 21st Century applications, the tools to manage them use technology from two decades ago. It is no wonder that a recent poll found that most IT shops first learn about incidents when users call their help desks. Essentially, they have no reliable visibility into their infrastructure environments, let alone the ability to forecast problems before users notice them. Now with the growing adoption of BSM and virtualization, the complexity is accelerating beyond the breaking point. Something has to change. Trying to manage systems or business services with conventional monitoring tools alone means you are constantly inundated with more data than you can possibly analyze and act on. Used in cojunction with Netuitive, however, the value of tools such as Tivoli can be unlocked and increased as the raw data they provide is used to manage enterprise estates more efficiently and more cost effectively.

By applying advanced contextual analysis -- using multivariate regression --and other math-based techniques, Netuitive stands on its own as the industry's only true self-learning performance management software. As a result, Netuitive self-learning performance management software delivers one-of-a-kind benefits, including:

• Automates performance management administration - eliminates rules, scripts and thresholds.
• Isolates root-causes across IT silos (physical and virtual).
• Forecasts issues up to two hours ahead.
• Provides unprecedented visibility of system and service performance, including composite health indexes.
• Provides an an intelligence layer on top of existing monitoring solutions with no new agents to deploy.

Only through a true self-learning approach is a solution like Netuitive's able to deliver accurate views of Service Health, Adaptive Behavior Profiles, Trusted Alarms and Forecasting.

Want to know more?

Orb Data is excited by the potential of Netuitive's products, and as a result has become their UK Services partner. For further information on how Orb Data and Netuitive can help you evolve your Performance Management Solutions, please contact This e-mail address is being protected from spambots. You need JavaScript enabled to view it