Choosing an Infrastructure Monitoring Tool

When choosing an infrastructure monitoring tool, consider one that offers the following features:

• Cloud-native and autoscaling support

  If you utilize serverless functions, containers, or cloud services in your stack, you’ll need to use an infrastructure monitoring tool that integrates with third-party cloud providers and orchestration tools. To track ephemeral, autoscaling cloud resources, some infrastructure tools will automatically start collecting data from backend components as they come online.

• Tagged infrastructure metrics

  Some infrastructure monitoring platforms can automatically tag backend components with applicable metadata—such as the operating system or service it’s running, the cloud provider, or the availability zone in which a host is located. These tags allow developers to aggregate metrics from across their infrastructure and target parts of their stack—such as a specific service or category of customers—that may be experiencing issues.

• Customizable alerts

  Infrastructure monitoring tools typically alert you when a key metric goes above or below a threshold. Some platforms also let you set up proactive, machine learning-based alerts that notify the appropriate teams when the error rate or latency of a host or container is trending higher than normal.

• Visualizations

  Some infrastructure monitoring tools provide pre-built or customizable dashboards that give you an overview of the health and performance of your hosts and containers. You can use these visualizations to identify overloaded hosts that require more resources or idle hosts that can be migrated to smaller instance types or consolidated to save on compute costs.

• Machine learning

  Machine learning-based tools automatically analyze historical infrastructure performance to detect anomalies, so you don’t have to manually set up alerts for every possible failure mode. For example, an infrastructure monitoring platform may automatically alert you if there’s an unexpected decrease in database query throughput. Machine learning-based tools may also forecast where CPU usage, memory usage, and other resource metrics are heading based on historical analysis.

• Unified platform

  A unified infrastructure monitoring platform gives you the ability to correlate infrastructure metrics with related traces, logs, processes, and events. This provides the full context of a request and allows developers to quickly diagnose and solve problems.

Datadog Infrastructure Monitoring offers full visibility into infrastructure performance across any on-premise, hybrid, or cloud environment. The easy-to-deploy, open source Datadog Agent collects metrics from your hosts and containers at 15-second granularity, and turn-key integrations with more than 500 popular technologies ensure complete coverage of your environment.

With tag-based search and analytics, you can slice and dice your metrics to create fine-grained alerts or to focus investigations on a specific subset of your infrastructure. Datadog also provides machine learning-based tools to proactively detect issues for you. Datadog’s unified platform brings together infrastructure monitoring with application performance monitoring, log management, digital experience monitoring, and more—giving you everything you need to understand and resolve performance issues across any layer of your stack.