Course contentsShow
← Cloud and Infrastructure
Lesson 629 of 2,98912. Monitoring FoundationsFree lesson

Time Series Data Fundamentals

How metrics are stored as time series with timestamps, values, and dimensional labels.

Time Series Data Fundamentals

You'll learn: How monitoring systems organize metrics as time-stamped data points with labels to track changes over time.

What Is Time Series Data?

Time series data is simply measurements recorded at specific moments in time. In monitoring, every metric you collect—CPU usage, request count, memory consumption—becomes a sequence of timestamps + values stored chronologically.

Think of it like a continuous temperature reading from your smart thermostat: [2024-01-15 10:00:00, 68°F], [2024-01-15 10:05:00, 69°F], [2024-01-15 10:10:00, 70°F]. Each reading is a data point in your temperature time series.

The Three Components

Every time series data point has three essential parts:

1. Timestamp — The exact moment the measurement was taken (usually in Unix epoch or ISO 8601 format)

2. Value — The actual measurement (could be a counter, gauge, histogram bucket, etc.)

3. Dimensional Labels — Key-value pairs that add context and allow filtering

For example, a CPU metric might look like:

timestamp: 1705324800
value: 75.3
labels: {
  host: "web-server-01",
  region: "us-east-1",
  environment: "production"
}

Why Labels Matter

Labels transform a simple number into queryable, filterable data. Instead of one giant "CPU usage" metric, you can slice by server, region, or environment—asking questions like "What's the CPU on production servers in us-east-1?"

This dimensional approach is what makes modern monitoring powerful. The same metric name with different label combinations creates unique time series streams.

Key Takeaway: Time series data stores metrics as timestamped values plus dimensional labels, enabling you to track measurements over time and filter by context like server, region, or application.