Monitoring Tools Profile: IBM Tivoli Monitoring

Always at or near the top of the big leagues in server management and monitoring, IBM Tivoli Monitoring represents an ongoing effort to harness enterprise-level monitoring into a unified framework (Tivoli Management Framework). Tivoli Monitoring 5.1 combines the capability of two previous products: IBM Tivoli Distributed Monitoring and IBM Tivoli Web Component Manager. It continues to develop the advantages of using a Web-based (IP) approach, such as providing a single point of access through the Tivoli Monitoring health console, and it enhances IBM’s push into proactive monitoring and repair capability.

When it comes to server management and monitoring, IBM Tivoli Monitoring invariably comes in at or near the top of the big leagues. The product suite is Big Blue’s ongoing effort to harness enterprise-level monitoring into a unified framework.

Tivoli Monitoring is actually a line of products within a large suite of management software that IBM sells under the Tivoli brand. Although the Tivoli Monitoring products can operate independently of the larger suite, they are intended to work with and complement those products. The key product in the Tivoli Monitoring line is Tivoli Monitoring, which provides the basic software. Within that, several products are targeted at specific areas of monitoring. For example, Tivoli Products, A Sampling

There is also Tivoli Monitoring for Network Performance, which is a somewhat different animal and is aimed at monitoring and optimizing network and internet transaction performance.

Important to bear in mind is that Tivoli NetView, a product not in the Tivoli Monitoring line, provides in-depth network monitoring, and several of the Tivoli Monitoring products require Tivoli Framework. End-to-end, the Tivoli line-up is vast enough to require a sufficient amount of time be set aside to determine which products apply to a specific situation, and an IBM consultant or sales representative is a reliable source for help steering the course.

Part of the task in fitting Tivoli products to an organization must also be based on the requirements of the software architecture. This consists of: Tivoli region management servers running basic Tivoli Monitoring software and potentially other Tivoli management programs (e.g., Tivoli Data Warehouse and Tivoli Business Systems Manager); Tivoli gateway/node software that aggregates and routes monitoring traffic as well as operates the heartbeat function (a scheduled poll of vital resource status); and Tivoli endpoints, which are typically the physical devices (servers and operating systems) being monitored. Although Tivoli provides excellent tools for the distribution and deployment of the software, careful selection of software placement and configuration is crucial.

Tivoli Monitoring is largely a Java-based application (and thus requires the Java runtime environment). It runs on a wide variety of operating system platforms that support Java.

The included Web Health Console provides administrative access to Tivoli Monitoring. It highlights the strengths of a Web-based approach, especially single log-on from anywhere in the world and a standardized graphical user interface (via a Web browser). In the larger Tivoli context, Tivoli Monitoring can also be administered via the Tivoli Enterprise Console and the Tivoli Business Systems Manager.

One of Tivoli Monitoring’s most distinguishing features is IBM’s ongoing quest to go beyond simple monitoring to provide automatic (or, in IBM parlance, “autonomic”) analysis and repair. The heart of the approach is resource modeling, which has two aspects, a dynamic model and a reference model.

The dynamic model is essentially a description of a resource object (e.g., a messaging server). Using the Common Information Model (CIM) or Microsoft’s variant WIM, it is an aggregation of physical properties (such as memory capacity and CPU utilization) that can be monitored on a given system. Most monitoring occurs through Simple Network Management Protocol (SNMP) or Windows Management Instrumentation (WMI) capabilities.

Once a resource object is identified and described, a reference model can be created that specifies the rules of operation for that object. This includes operational logging, expected levels of performance, baseline metrics, and signatures of various problem situations. The reference model makes it possible for Tivoli to analyze relatively complex operational problems and take action — either by sending alerts or attempting automated intervention. A collection of resource models can be assembled into a profile, which provides a perspective on those resources. Multiple profiles can be developed to look at the system operations from different angles.

Taking the reference model further, IBM uses Proactive Analysis Components (PAC) to identify problem signatures, often a combination of individual monitoring alarms. It provides the means to deal with time and frequency related problems, for example a cache usage rate that fluctuates between overload and normal during a specific time period. By combining performance logs with known capacities and expected metrics, Tivoli Monitoring can be trained (in a sense, programmed) to differentiate between a one-time resource problem and a persistent bottleneck.

Tivoli Monitoring products ship with a number of typical resource models. The Tivoli Monitoring Resource Model Builder (formerly Tivoli Monitoring Workbench), for example, adds the capability to edit and create customized resource models.

With its sophisticated capabilities, Tivoli Monitoring is likely to be most efficient (and appreciated) where large number of resources must be monitored and where IT expertise to configure, update, and refine the analytical and response features is present. Tivoli Monitoring basically demands considerable resources be expended to organize, plan, implement, and maintain the monitoring system.

Developing a monitoring system that effectively crosses operating systems, LANs, WANs, and IP domains for potentially thousands of devices and applications is no mean feat. IBM is among only a handful of vendors (including HP, BMC, and Computer Associates) that has the wherewithal and enterprise reach to pull it off. Even so, it has taken IBM years to bring the Tivoli products closer to an integrated format. Thanks to the Web and IP, a degree of consolidation and unification has been possible that wasn’t before, and Tivoli Management 5.1 is a good indication of the direction server system monitoring is taking.

User support for Tivoli is available by telephone, and on-site, as well as through training, user groups, online forums, IBM’s knowledge base, newsletters, developer domain, and e-mail.

Server Monitoring Tools Matrix
Tivoli Monitoring