Most observability solutions claim to address a broad range of business use cases, but architectural compromises limit success. To overcome these compromises, Dynatrace recently introduced business events powered by Grail. Metric extraction extends the value of business events, adding new flexibility and greater simplicity to Dynatrace business events and analytics.
Should business data be part of your observability solution?
Technology and business leaders express increasing interest in integrating business data into their IT observability strategies, citing the value of effective collaboration between business and IT. Exploding volumes of business data promise great potential; real-time business insights and exploratory analytics can support agile investment decisions and automation driven by a shared view of measurable business goals.
But are observability platforms—born from the collision between the demands of cloud computing and the limitations of APM and infrastructure monitoring—the best solution for managing business analytics?
Observability fault lines
The monitoring of complex and dynamic IT systems includes real-time analysis of baselines, trends, and anomalies. This is achieved, in part, by establishing actionable statistical accuracy—not necessarily precise accuracy—through practical levels of metric sampling, aggregation, and extrapolation. Statistical accuracy is typically not precise enough for most business use cases, and business actions can’t be automated based on anything less than 100% precise data.
Similarly, many performance analytics goals are served by capturing method calls and arguments to give developers the information they need to diagnose and correct anomalies. Traditional observability solutions don’t capture or analyze application payloads. Such analysis is intentionally excluded from most observability solutions because payload details are unnecessary for DevOps purposes, problematic for agent overhead, and risky for data privacy. Therefore, to add the “Biz” to “DevOps,” observability solutions typically rely on application log files, which incurs significant development and maintenance overhead.
To close these critical gaps, Dynatrace has defined a new class of events called business events. Dynatrace OneAgent® prioritizes business events over observability metrics to ensure the lossless precision you need to support demanding business use cases. At the same time, deep payload inspection makes it easy to extract important business data locked in application payloads—without writing any code. When combined with Dynatrace Grail™, business events gain long-term, cost-effective storage, unaggregated granularity, instant indexless queries, and automatic Smartscape® topology context.
Introducing metric extraction from business events
Beginning with Dynatrace SaaS version 1.257, you can extract metrics from ingested business events. Metric extraction is a convenient way to create your business metrics, delivering fast, flexible, and cost-effective analytics. Since these metrics are derived from business events, they inherit key characteristics, including lossless precision and deep access to application payloads. Extracting metrics from business events allows you to use metric expressions, view baselines and trends, and create custom alerts when issues require further evaluation by Dynatrace Davis®, our causational AI engine.
From business events to metrics: the ingest pipeline
All incoming business events must pass through the business event ingest pipeline where transformational processing occurs. It’s at this point—immediately after processing—where you can extract metrics from your events by defining extraction rules.
Extraction rules include the following four fields:
- Key: Always starts with the prefix
- Matcher: A matcher-specific DQL query that selects specific incoming business events
- Measure: Choose an event-attribute value or an event-count value
- Dimension: Define up to 50 dimensions
In this example, we use a DQL matcher query to extract the
trading_volume attribute value from all incoming business events that have an event.type of
com.easytrade.buy-assets. We assign the metric key of
bizevents.EasyTrade.TradingDollarVolume, and include the
dt.entity.host dimension, which will allow us to split or filter the results by host.
Once defined, custom business-event metrics will be extracted from all future ingested business events that match this rule.
View and analyze
You can use the interactive Data explorer to view, query, transform, and visualize your metrics. Once you’ve refined the visualization of a metric, you can pin it to a dashboard.
Through filtering, aggregation, or arithmetic operations, Data explorer also supports metric expressions that combine one or more existing metrics to create new, on-demand metrics. For example, you could divide the
TradingDollarVolume metric by a
TradeCount metric to calculate a new metric,
AverageTradeValue. The new metric can then be pinned to a dashboard. The Data explorer supports many types of visualizations, making it easy to understand trends or to view the relationship between business and IT metrics.
Metric events for alerting
Metric event configuration can be used to automatically detect anomalies in metric timeseries using thresholds or baselines. The resulting alerts can be based on attribute values or specific event occurrences. The extraction of metrics from business events automatically preserves the Smartscape context of each event, allowing Davis AI to automatically analyze anomalies and identify any relevant impact and root cause.
Conclusion: Metric extraction adds new value to business events
Lossless precision and easy, code-free access to business data distinguish business events from traditional IT metrics. Powered by Grail, business events gain long-term cost-effective storage, unaggregated granularity, instant indexless queries, and automatic Smartscape context.
Extracting metrics from business events delivers additional value, including comprehensive metric calculations, a wide range of visualizations, threshold-based alerting, and AI-driven causation analytics.
Metric extraction from business events gives you new options for monitoring, analyzing, and automating your business use cases. For additional technical insights on extracting metrics from business events, watch this Business Events Observability Clinic.