Alert Lens

Focus on what’s important
Alert Lens is an intelligent monitoring solution that helps IT Operations improve up time with less alert fatigue. Bolt’s AI and ML univariate, multivariate, supervised and unsupervised models deliver prediction, detection, correlation, and root cause faster, and with greater accuracy to identify and fix anomalies.

A broad set of data sources is supported with additional sources easily added, and anomalies are sent to a variety of notification tools. All alerts are maintained in a searchable database allowing users to review the alert history and drill down for deeper analysis on the activities that created the alert.

Alert Lens combines existing monitoring data from multiple data sources then uses real time AI/ML models to:

Detect anomalies in metric streams to create accurate events.
Correlate events from across multiple data sources.
Quickly zero in on the root cause.

Architecture

Built as a set of containers, Alert Lens uses TensorFlow as its fundamental building block for the neural models. To allow for fast ingestion and inference of metrics data, the Kafka platform is used along with a purpose built monitoring analytics pipeline. This combination provides for fast and scalable alert processing that can easily be extended to support new data connectors and additional data processing stages.

Purpose-built for the Analysis of Time Series Data

Alert Lens is built on a highly scalable data science platform purpose-built for the analysis of time series data. The real-time univariate, multivariate, and correlation AI/ML models support a wide variety of time series data and are highly adaptive both in initial deployment and over time.

Supervised and unsupervised training optimizes to deliver accurate inferences, and automatic retraining adapts to changes in the environment and data sources for improved results and reduced false positives.

Alert Lens can scale to tens of thousands of time series KPIs running concurrently, without the need for expensive GPU-based infrastructures, yet still deliver results in real time. Alerts can be delivered via notification tools such as PagerDuty or Slack, or can be used programmatically using Kafka topics or AWS SNS.