SignalFx Developers Guide

Data Model Basics

SignalFx monitoring is based on the SignalFx data model, in which data consists of:

  • A numerical measurement called a metric

  • A metric type

  • One or more dimensions

  • Dimensions metadata that includes custom properties and tags.

Each piece of data in this form is called a datapoint. The set of datapoints that have the same metric name, metric type, and set of dimensions form a metric time series, usually abbreviated as MTS.


A metric is a numerical measurement that changes over time. Some metrics measure hardware and infrastructure parameters:

  • Percentage use of CPU per core

  • Network throughput

  • Memory use percentage

Other metrics measure application and service parameters:

  • Number of transactions processed per second

  • Percent uptime

  • Number of running instances


Regardless of the measurement, the SignalFx data model treats incoming metrics as datapoints that contain the following properties:

  • Metric name: For example, cpu.utilization.

  • Value: For example, .92

  • Metric type:

    • Gauge: Reports the value of some measure at a point in time. The percent utilization of a cpu core is an example of a gauge metric.

    • Counter: Reports the number of occurrences of an event or activity over a period of time. The number of pages send by a web server in a 60-second interval is an example of a counter metric.

    • Cumulative counter: Reports total activity over the lifetime of an application or process. The number of API calls received by a server since the server started up is an example of a cumulative counter.

  • Array of properties called dimensions. A dimension doesn’t have to be numeric, and usually doesn’t change over time. A server’s hostname is an example of a dimension.

  • Optional timestamp. If no timestamp is provided, SignalFx sets the timestamp to the time at which it receives the datapoint.

Metric time series (MTS)

A set of datapoints that all have the same metric name and list of dimensions is known as a metric time series (MTS). Since statistical calculations are based on one or incoming MTS, and charts and detectors are based on statistical calculations, MTS are the heart of SignalFx functionality.

© Copyright 2020 Splunk, Inc.

Third-party license information