Instrumentación y métricas clave en performance para aplicaciones Node.js

Understanding how instrumentation impacts application performance in Node.js is crucial to optimize resources and improve user experience. Instrumentation allows collecting information about the performance of a program by capturing internal code events. Although useful, this process comes at a cost in terms of performance, as additional code execution can reduce the main performance of the application.

What is instrumentation and how does it impact performance in Node.js?

Instrumentation in Node.js is a technique for listening to events in the code and obtaining data about its performance. However, when implementing this technique, additional code is inserted that consumes resources and can negatively affect overall performance. When measuring the number of requests handled per second, different instrumentation tools can vary in their impact on the overall capacity of the application.

For example, without instrumentation, Fastify handles about 63,000 requests per minute. By enabling observability with Solid from NodeSource, the number decreases by approximately 3,000 requests per minute. If we incorporatetracing, it decreases by almost 9,000 requests. Other providers such as Datadog or Dynatrace can impact performance even more drastically, dropping to 15,000 or even 5,000 requests respectively.

What are the essential metrics to evaluate Node.js performance?

Properly evaluating an application in Node.js requires specific attention to several types of metrics. These are categorized into:

System metrics and basic information.

• Number of CPU cores.
• CPU model and architecture.
• Total memory, free memory and system load.
• System uptime.

Specific metrics for Node.js processes.

• CPU usage by the process.
• Processuptime.
• Total resident memory (RSS).
• Heap usage (memory available to instantiate objects).
• External memory usage (for native C++ modules).

Critical Event Loop Metrics

• Average event loop execution time.
• Estimated delays(lag) in the event loop.
• Event loop occupancy level(event loop utilization), considered by many as the most critical metric in Node.

Garbage Collection Metrics

• Frequency and duration of garbage collection (garbage collector) for effective memory management.

Additional metrics according to functionalities used

• HTTP statistics: number and duration of requests.
• I/O primitives: network consumption, reads, writes to disk.

How to effectively monitor a Node.js application in different environments?

There are two main contexts for monitoring Node.js applications:

Local monitoring (development environment).

It is advisable to use specific tools designed for performance testing and diagnostics in development:

• Clinic.js (Clinic Doctor): Specialized tool for analyzing metrics concerning CPU and memory through an encapsulated process that records metrics while applying load tests.
• Autocanon: Used in conjunction with Clinic.js for simulation and user load.
• 0x: Specific tool for examining CPU usage locally, accessible via simple commands such as npx 0x.

These tools are mainly designed for development analysis, they are not recommended in production due to the additional load and control that the infrastructure would need.

Monitoring in production

In production, it is crucial to use tools that provide real-time information on vital metrics. System performance management applications (APMs) allow continuous evaluation of factors such as CPU, memory and network activity.

• NodeSource's EnSolid: An open source runtime designed specifically for Node.js that provides accurate real-time metrics, ideal for both local and production use, ensuring full compatibility and minimal impact on the observed application.

Determining the right monitoring tool according to the current environment (local or production) is an essential strategic decision to efficiently manage resources and optimize performance in Node.js. Careful consideration should always be given to what style of observability is needed based on the specific conditions of the project.