Agent Deployment Guidelines

When planning a deployment, developing the balance between agent overhead and the visibility of application performance is key. Low overhead agents allow for real-time transaction monitoring in production environments. Low-overhead agents help performance and availability of critical applications and server resources. 
apmdevops106
When planning a deployment, developing the balance between agent overhead and the visibility of application performance is key. Low overhead agents allow for real-time transaction monitoring in production environments. Low-overhead agents help performance and availability of critical applications and server resources. 
However, at times, keeping overhead low may not provide enough information to diagnose problems when they occur. Often Administrators deploy and fine-tune the agent configuration throughout the application lifecycle. Administrators often monitor more components when an application is being developed or tested, then reduce the components after releasing an application to production.
Install and Evaluate the Default Functionality
  • Install and evaluate the default agent configuration that provides data collection for common components of the application and the computing environment.  
  • As you evaluate the depth and breadth of data collection, you become familiar with Introscope. When you are comfortable monitoring applications, you can customize the agent data collection. Keep in mind that when the agent collects more metrics, it consumes more system resources. If the agent collects fewer metrics, you have less visibility into potential problems. 
  • Refine the agent configuration. Continue to evaluate the balance between the depth of data collection and an acceptable performance level. The appropriate level of instrumentation typically depends on where the agent is deployed. For example, an agent monitoring within a test environment is typically configured to collect large numbers of metrics. An agent on a production server, however, is typically configured to deliver essential information.
Determine Configuration Requirements through the Application Lifecycle
  • Determine your data collection requirements. You can vary the data collection behaviors of the agent and can evaluate the impact on overhead through alternative configurations of the agent.
  • Configure the agent to match the monitoring goals, environmental constraints, and service level expectations for each stage of the application lifecycle. Also configure the agent based on the type of environment you are monitoring. For example, in pre-production application environments, you typically configure a higher level of data collection. This level provides deeper visibility into the application performance. In production or high-volume transaction environments, you typically reduce the metrics reported to control agent overhead. 
  • Determine the appropriate trade-off between the visibility of performance details and resource overhead. Consider the optimum level of visibility at a reasonable cost for the environment you are monitoring. Depending on your requirements, you can also configure agent properties to control specific agent behavior. For example, track the maximum number of metrics that are collected and the removal of old metrics.
Define a Baseline Agent Profile with Specific Configuration Properties
  • Create a "candidate" agent configuration. Most agent operations are controlled using properties in the IntroscopeAgent.profile file, also named the
    agent profile
    . For example, the IntroscopeAgent.profile file defines 
    ProbeBuilder Directive
     and 
    ProbeBuilder List
     files. The agent uses these files, which control the specific metrics that the agent gathers. The IntroscopeAgent.profile file also provides properties that enable specific features to fine-tune operations, such as polling intervals.
  • Adjust the properties in the agent profile to evaluate the impact of each change. For example, start with a default agent profile that does not have ChangeDetector enabled. Later, you can enable ChangeDetector properties in the profile and can evaluate agent performance before adding other features.
Evaluate Agent Performance Overhead
  • Verify that the metrics that are collected provide sufficient visibility into application performance and availability. In addition, verify that the volume of metrics cannot impose an unacceptable load on the operating environment. Do not configure the agent to report more metrics than are necessary to identify and localize performance and availability problems.
  • Evaluate agent performance by understanding the performance characteristics of the application. For example, you can load test your application before and after implementing default monitoring to verify impact.
  • Extend data collection in a controlled way by verifying agent operation and overhead before and after implementing changes. For example, only add monitoring support for one application at a time to allow you to evaluate the performance of each add-on individually.
Validate and Deploy the Agent Configuration
  • Deploy a validated configuration by installing the IntroscopeAgent.profile file and modified or custom PBD files to the target environment.