Rolling Review: Quest Software Foglight

Lots of APM vendors talk about taking a holistic approach to application performance management, but the latest entry in our Rolling Review, Quest Software's Foglight, delivers. This product combines network packet capture data, in-depth analysis using intelligent agents and synthetic transactions to enable APM. The system comprises the Foglight Experience Monitor, or FxM; the Foglight Experience Viewer, or FxV; and a new implementation of the Foglight Server to offer views ranging from business services to those low-level traces only an application developer could love.

Foglight's greatest strength is its ability to collect many different types of data and present it all in a meaningful way. There are literally thousands of metrics that are baselined and processed via the rules engine. Alarms may be generated when thresholds are crossed or rules are created. These alarms can be sent via several methods, including SNMP traps, or collected into dashboards. Foglight dashboards can be created easily and set to model high-level business and service views that clearly identify issues as they occur. Drill-down capabilities provide access to lower-level statistics, useful for troubleshooting. FxV offers user-session replay, click by click, for Web applications. This is incredibly valuable for diagnosing problems, as it illustrates the exact order of events that caused a performance glitch.

All that data does come with a price in terms of complexity. Still, though Foglight is not as plug-and-play as some of the other tools we've reviewed, in particular those from Indicative or Nimsoft, most IT organizations should be able to deal with the application as long as they have an effective deployment plan. Quest or third-party systems integrators are more than happy to help here, and while services costs may rival that of the actual software, the expense should be worth it.

The virtually infinite flexibility and vast amount of data collected by the various monitors can be overwhelming. To help here, Foglight Server uses models to represent services and applications. Models are the key to the system's flexibility and power because they group and assign dependencies to all metrics collected. As in other APM systems, you will require a pair of hands and some knowledge to choose the applications and services that will be modeled, then break them down into component pieces. Applicable metrics are identified along with their dependencies as each model is built. Knowledge of each component is essential, but many of the cartridges include default metrics and thresholds as a starting point.

Application Performance Optimization Immersion Center

NEWS | REVIEWS | BLOGS | FORUMS
TUTORIALS | STRATEGY | MORE

Given the depth of metrics collected, implementation of models is easier with Foglight than in most products we've seen. Development of models centers around a standard tree navigator. Metrics are grouped in several ways to expedite locating them. Choosing a plus or minus either adds or removes the metric from a model definition.

The Great Agent Debate

For deep application performance monitoring, there's no getting around it: Agents are critical. We've found that products with agents provide more detailed and in-depth statistics on application-level processes and data, all crucial in troubleshooting and diagnosing application problems.

Quest's Foglight Cartridges are deployable application-specific units that include monitoring agents, metrics, database modification scripts and rules-all the components with the necessary intelligence for monitoringvital applications or databases. A wide variety of Cartridges are loaded by default, and Foglight support most leading business apps, including Java, .NET, SAP, Oracle E-Business Suite, PeopleSoft, Siebel, Oracle Database, Microsoft SQL Server, IBM DB2, Sybase and IBM WebSphere MQ. Note that some Foglight 4.x Cartridges, including those for IIS and ASP/.NET, have not been ported to the new version 5 architecture. The ASP/.NET Cartridge is scheduled for release by year's end, and the IIS Cartridge is currently due in Q1 2008.

Configuration of agents, such as a watched-process list, can be assigned via templates, so it's fairly easy to manage a collection of agents. It would be nice to see more grouping and group functions, rather than an agent-by-agent approach to configuration. However, much of the configuration can be done via a command-line utility, useful for performing repetitive tasks.If Agents Are a Showstopper ...

While many organizations cringe at installing and maintaining more agents, for true application-specific data they are a necessity, and Quest's approach is a reasonable one.

Still, if you just can't bring yourself to add one more agent, Quest can provide visibility into your critical applications anyway, albeit without as much depth. Synthetic transactions are created using a Recorder that is a wrapper around an Internet Explorer browser object; this makes their creation easier, and being able to tweak each step of the transaction enables simple troubleshooting and resolution of issues.

In addition, Quest offers an appliance that focuses on the collection of application packets across the network. The appliance deployment was straightforward and involved configuring the network settings in a typical Linux text menu. Integrating appliance configuration into the Foglight Server is on the product roadmap, and it will eliminate the piecemeal approach that is prevalent today.

Michael Biddick is the executive VP of Solutions for Windward Consulting Group, a firm that helps organizations improve it operational efficiency, and contributing editor for Network Computing/InformationWeek. Write to him at [email protected].