Falkonry Time Series Intelligence¶
Time Series Intelligence (TSI) is an AI-native software tool designed to analyze real-time metrics from telemetry and instrumentation sources with minimal programming effort. It enables engineers to extract actionable insights from observability data across industrial, IT, and networking systems. By efficiently processing large volumes of data, TSI identifies and highlights periods and subsets of time series containing the most informative data. This facilitates alerting, root cause analysis, action planning, and comparative assessments to drive continuous operational improvement.
Key Features¶
Observability systems generate vast amounts of complex data that traditional techniques like business intelligence, complex event processing, search, or mathematical modeling struggle to handle effectively. Deriving actionable insights from such data requires advanced methodologies, including stream processing, signal processing, convolutional analysis, and logical reasoning.
Features Overview¶
Time Series Intelligence addresses the challenges of processing observability data and offers the following AI-driven capabilities:
- No-code/low-code exploratory data analysis: Simplifies data exploration without extensive programming expertise.
- Time series embedding generator: Captures essential characteristics of time series data for downstream tasks.
- Time series anomaly detection and classification: Identifies and categorizes deviations from expected behavior.
- Machine learning model management: Organizes, deploys, and monitors machine learning models.
- Model inference scheduling: Automates continuous and efficient model predictions.
- Dependency management for time series calculations: Handles interdependent time series transformations efficiently.
- Data aggregation for scaled processing: Combines large datasets for streamlined analysis.
- Metadata-based dynamic dashboards: Creates dashboards that adapt based on metadata attributes.
- Visualization tools for time series: Offers dynamic and customizable visual representations of time series data.
Capabilities¶
TSI empowers engineers to interrogate and derive inferences from time series data.
Data Ingestion¶
TSI ingests data into a streaming interface, enabling seamless processing and analysis. Data can be ingested from a variety of sources while dealing with data flow challenges resulting from upstream errors, networking glitches and changing operations.
Reports¶
Reports visualize time series data through flexible formats such as time plots, value distribution plots, and temporal trends. They provide dynamic access to time series data based on time ranges, enhancing visibility and root cause analysis.
Calculations¶
Calculations transform raw data into derivative time series, highlighting domain-specific characteristics based on expert knowledge. These transformations, akin to Matlab expressions and Python functions, offer a flexible and intuitive mechanism for applying standard signal derivations.
Rules¶
Rules convert analog quantities like raw signals, anomaly scores, and condition labels into discrete events based on thresholds automatically assessed by Falkonry TSI. They enable condition-based actions through a no-code interface, leveraging specific behaviors derived from signal combinations.
Insights¶
Insights automatically analyze time series data for anomalies by identifying deviations from normal operating behavior without requiring setup or supervision. They help users understand the origins and contributors of anomalies, enabling appropriate actions to mitigate their effects and likelihood.
Patterns¶
Patterns uncover meaningful temporal correlations, such as early warnings or stages of deterioration, in complex systems using multivariate data streams. Designed for reliability engineers and data analysts, Patterns facilitate continuous monitoring of specific events and known conditions, as well as root cause analysis to identify permanent fixes.
To quickly start processing your time series data through TSI, please see our Quick Start Tutorial.