AppDynamics, now part of Cisco, is an application performance monitoring platform aimed at large enterprise applications, especially Java and .NET estates with deep transaction chains. Its organizing concept is the business transaction: rather than monitoring raw endpoints, it groups requests by the business operation they perform - "Checkout", "Login", "Fund Transfer" - and tracks the health of each as a unit, end to end across every tier it touches. This framing lets an operations team report on the systems in terms a business owner recognizes, not just in HTTP routes.

Business transactions and the flow map

An agent instruments the application and automatically detects entry points, classifying each request into a business transaction by URL pattern, servlet, or custom rule. AppDynamics then draws a flow map of the application's tiers - web, app, cache, database - and measures the response time and error rate of each business transaction as it crosses them. When a transaction slows, the platform shows which tier added the latency and, through snapshots, the exact call stack of the slow requests.

Health rules and baselines

Alerting is driven by health rules evaluated against dynamic baselines rather than fixed thresholds. AppDynamics learns the normal response-time and load pattern for each business transaction by time of day and week, then a health rule flags deviation from that learned baseline - so a transaction that is slow only relative to its own Monday-morning norm still triggers. A rule reads as a condition over a metric and its baseline:

Health Rule: Checkout latency degradation
  Affects: Business Transaction = /checkout
  Critical condition:
    95th percentile Response Time (ms)
      is GREATER THAN
      Baseline (Default) BY 3 standard deviations
    for 5 consecutive minutes
  Warning condition:
    Errors per Minute is GREATER THAN baseline BY 2 standard deviations

Frequently asked questions

What is a business transaction in AppDynamics? A logical grouping of requests by the operation they perform, tracked end to end across every application tier, so health is reported per business operation rather than per raw endpoint.

How does its alerting differ from threshold alerts? Health rules compare current behavior against a dynamic baseline learned per time of day and week, catching deviations from normal rather than only absolute limits.

What environments is it strongest in? Large, long-lived enterprise Java and .NET applications with many tiers, where deep automatic instrumentation and tier-by-tier flow mapping matter most.

How does it relate to Cisco? Cisco acquired AppDynamics, and it is now positioned alongside Cisco's full-stack observability portfolio for enterprise customers.

See the monitoring & logging overview and compare with Dynatrace.