Skip to content

性能和基准测试#

¥Performance and benchmarking

n8n 在单个实例上每秒最多可处理 220 个工作流执行，并且可以通过添加更多实例来进一步扩展。

¥n8n can handle up to 220 workflow executions per second on a single instance, with the ability to scale up further by adding more instances.

本文档概述了 n8n 的性能基准测试。它描述了影响性能的因素，并包含两个示例基准测试。

¥This document outlines n8n's performance benchmarking. It describes the factors that affect performance, and includes two example benchmarks.

性能因素#

¥Performance factors

n8n 的性能取决于以下因素：

¥The performance of n8n depends on factors including:

  • 工作流类型

¥The workflow type

  • n8n 可用的资源。

¥The resources available to n8n

  • 如何配置 n8n 的扩展选项

¥How you configure n8n's scaling options

运行你自己的节点基准测试#

¥Run your own benchmarking

要获得针对你用例的准确估算，请运行 n8n 的 基准测试框架。该仓库包含更多关于基准测试的信息。

¥To get an accurate estimate for your use case, run n8n's benchmarking framework. The repository contains more information about the benchmarking.

示例：单实例性能#

¥Example: Single instance performance

此测试衡量响应时间如何随每秒请求数的增加而增加。它查看调用 Webhook 触发节点时的响应时间。

¥This test measures how response time increases as requests per second increase. It looks at the response time when calling the Webhook Trigger node.

设置：

¥Setup:

  • 硬件：ECS c5a.large 实例（4GB 内存）

¥Hardware: ECS c5a.large instance (4GB RAM)

  • n8n 设置：单个 n8n 实例（在主模式下运行，使用 Postgres 数据库）

¥n8n setup: Single n8n instance (running in main mode, with Postgres database)

  • 工作流：Webhook 触发节点，编辑字段节点

¥Workflow: Webhook Trigger node, Edit Fields node

Graph showing n8n response times by requests per second

This graph shows the percentage of requests to the Webhook Trigger node getting a response within 100 seconds, and how that varies with load. Under higher loads n8n usually still processes the data, but takes over 100s to respond.

示例：多实例性能#

¥Example: Multi-instance performance

此测试衡量响应时间如何随每秒请求数的增加而增加。它查看调用 Webhook 触发节点时的响应时间。

¥This test measures how response time increases as requests per second increase. It looks at the response time when calling the Webhook Trigger node.

设置：

¥Setup:

  • 硬件：七个 ECS c5a.4xlarge 实例（每个实例 8GB 内存）

¥Hardware: seven ECS c5a.4xlarge instances (8GB RAM each)

  • n8n 设置：两个 Webhook 实例、四个工作实例、一个数据库实例（MySQL）、一个运行 n8n 和 Redis 的主实例。

¥n8n setup: two webhook instances, four worker instances, one database instance (MySQL), one main instance running n8n and Redis

  • 工作流：Webhook 触发节点，编辑字段节点

¥Workflow: Webhook Trigger node, Edit Fields node

¥Multi-instance setups use Queue mode

Graph showing n8n response times by requests per second

This graph shows the percentage of requests to the Webhook Trigger node getting a response within 100 seconds, and how that varies with load. Under higher loads n8n usually still processes the data, but takes over 100s to respond.