APM Rolling Review: ProactiveNet

In June, BMC Software acquired ProactiveNet, and we tested the product for our application performance management Rolling Review. ProactiveNet isn't usually considered a player in the APM market, as it's focused on service-level agreements. However, ProactiveNet will notify IT of impending problems it finds by applying algorithms to stored data. The app shines as an intelligent correlation tool, unlike other products tested that focus on data gathering.

If you require collection of network application traffic or synthetic transactions, ProactiveNet is not for you. However, it does offer agents that can be deployed to assemble information, or you may use its adapters to pull data from CA, BMC Performance Manager, or NetIQ. Agents are available for application components that range from WebLogic and WebSphere to infrastructure components such as Oracle, Unix, AIX, IIS, Apache, and dozens of others.

DATA ENTRY

ProactiveNet doesn't autodiscover your environment; you'll need to use its agents or another enterprise management adapter to get data into the system. The downside is that if a monitor doesn't exist for your application, you'll have to create a custom monitor that intensively queries the target server for available metrics. Metrics to be monitored must be selected and assigned names. To help speed the process, ProactiveNet provides template functionality for Oracle, WebSphere, WebLogic, and custom monitors, creation of which requires some design and metric definition.

Once setup is done, you then need to establish relationships within devices or device classes. Like devices, relationships must be defined--they're not discovered. Once you specify and group underlying infrastructure components that comprise application services, relationships will determine which alarms will be suppressed and which paths root cause analysis will follow.The ability to create service groupings is powerful, but isn't as flexible as we'd like. For example, the device description field is populated with the location property. All devices with the same description field are grouped under that location property name. Views can't easily combine multiple properties, which would be useful for complex service reporting. Reports also can't be grouped by property; we couldn't group our devices by location, customer name, and criticality.

Once SLAs are configured and relationships established, capacity trend reports can be customized to predict SLA violations. We really like this ability--it's nice to get early warning so you actually have a chance to forestall a problem. Both ProactiveNet server and its agents have native failover capabilities, critical for larger enterprises. ProactiveNet also can forward SNMP traps.

In our environment, ProactiveNet agents were able to detect performance problems within our application that were related to excessive CPU utilization. Agents reported this information back to the console, and we were alerted that our SLA had been violated. These real-time warnings will help prevent critical outages.

ProactiveNet sends alerts by pager or e-mail for problem areas that may impact application performance or violate SLAs. If you don't want to be woken up, you can create automatic triggers with corrective actions for alarms that meet specific criteria. When response time for an application slows, ProactiveNet will seek to determine the cause of the problem. Even better, it will score and group events in real time to identify infrastructure components responsible for application performance degradation.

These relationships can be visualized across application infrastructure components, displaying as many as six attributes in a single graph. These graphs help identify trend variations, abnormalities, and other complex relationships, resulting in faster identification and resolution of performance issues.0