In Search of a Web Panacea

This is the benchmark needed to find the perfect programming language and web development framework at least the fastest.

Results

Machine: AMD FX-8320 4.0 GHz, 12 GB RAM.

get-tiny-json-entity

Name	Requests Total	Latency Average	2xx	Non 2xx	CPU % Average	Memory MB Average	Duration
Bare Node.js	2607953	2.24	2607953	0	_	_	60.55
Dragonrend.js	3545549	1.65	3545549	0	_	_	60.51
Express.js	2473805	2.35	2473805	0	_	_	60.47
Fastify.js	3026307	1.91	3026307	0	_	_	60.44
Koa.js	2930610	1.97	2930610	0	_	_	60.48
Java Http Server	2599748	2.23	2599748	0	249	462.7	60.43
Spring Boot	2186704	2.73	2186704	0	410	474.2	60.87
Proteus	2713172	2.17	2713172	0	101	259.5	60.6
Colossus	3482496	1.69	3482496	0	132	272.8	60.57
Blaze	3390464	1.72	3390464	0	116	212.8	60.54
Actix	3304113	1.77	3304113	0	97.5	11	60.59

get-large-json-entity

Name	Requests Total	Latency Average	2xx	Non 2xx	CPU % Average	Memory MB Average	Duration
Bare Node.js	2646970	2.19	2646970	0	_	_	60.49
Dragonrend.js	3513369	1.67	3513369	0	_	_	60.59
Express.js	2413458	2.4	2413458	0	_	_	60.44
Fastify.js	3033961	1.9	3033961	0	_	_	60.42
Koa.js	2886179	2	2886179	0	_	_	60.41
Java Http Server	2626802	2.21	2626802	0	256	432	60.49
Spring Boot	2175857	2.74	2175857	0	407	416.5	60.79
Proteus	2740137	2.15	2740137	0	104	256.2	60.59
Colossus	3486113	1.67	3486113	0	132	271.8	60.57
Blaze	3485128	1.68	3485128	0	121	211.2	60.56
Actix	3393272	1.72	3393272	0	96.5	11	60.54

post-tiny-json-entity

Name	Requests Total	Latency Average	2xx	Non 2xx	CPU % Average	Memory MB Average	Duration
Bare Node.js	2634937	2.2	2634937	0	_	_	60.45
Dragonrend.js	3321675	1.76	3321675	0	_	_	60.45
Express.js	2280709	2.55	2280709	0	_	_	60.51
Fastify.js	2890691	1.99	2890691	0	_	_	60.38
Koa.js	2672092	2.17	2672092	0	_	_	60.47
Java Http Server	2675910	2.16	2675910	0	214	502	60.37
Spring Boot	2060132	2.89	2060132	0	473	447.2	60.77
Actix	3822047	1.53	3822047	0	105	10.8	60.5

post-large-json-entity

Name	Requests Total	Latency Average	2xx	Non 2xx	CPU % Average	Memory MB Average	Duration
Bare Node.js	2595860	2.24	2595860	0	_	_	60.55
Dragonrend.js	3245853	1.81	3245853	0	_	_	60.59
Express.js	2289065	2.54	2289065	0	_	_	60.53
Fastify.js	2843923	2.03	2843923	0	_	_	60.42
Koa.js	2618795	2.21	2618795	0	_	_	60.5
Spring Boot	2076500	2.87	2076500	0	477	391.4	60.83
Actix	3896895	1.5	3896895	0	115	10.6	60.41

get-plain-text

Name	Requests Total	Latency Average	2xx	Non 2xx	CPU % Average	Memory MB Average	Duration
Bare Node.js	2652808	2.2	2652808	0	_	_	60.59
Dragonrend.js	3524767	1.67	3524767	0	_	_	60.58
Express.js	2514659	2.31	2514659	0	_	_	60.5
Fastify.js	3074613	1.87	3074613	0	_	_	60.39
Koa.js	2973095	1.94	2973095	0	_	_	60.44
Java Http Server	2687110	2.15	2687110	0	249	435.2	60.42
Spring Boot	2622008	2.26	2622008	0	420	445.9	60.69
Proteus	2806507	2.09	2806507	0	99.8	265	60.53
Colossus	3452516	1.7	3452516	0	132	415.3	60.61
Blaze	3416356	1.72	3416356	0	113	212.5	60.53
Actix	3501122	1.67	3501122	0	98.6	10.9	60.52

get-tiny-json-entity-by-id

Name	Requests Total	Latency Average	2xx	Non 2xx	CPU % Average	Memory MB Average	Duration
Dragonrend.js	3571999	1.64	3571999	0	_	_	60.64
Express.js	2512260	2.31	2512260	0	_	_	60.53
Fastify.js	3088123	1.87	3088123	0	_	_	60.41
Koa.js	2963847	1.94	2963847	0	_	_	60.38
Spring Boot	2138346	2.78	2138346	0	469	451.4	60.8
Proteus	3160474	1.86	3160474	0	113	268.8	60.68
Colossus	3615735	1.61	3615735	0	135	308.5	60.51
Actix	3810874	1.54	3810874	0	103	11.1	60.44

How to run?

If you want to run the benchmark yourself, then follow these instructions:

Requirements

Benchmark:

• Node.js and NPM

Subjects require languages support and build tools such as:

• Java, Maven
• Scala, SBT
• Rust, Cargo
• Node.js, NPM (should already be to start the benchmark)

Installation

$ git clone https://github.com/EgorRepnikov/in-search-of-a-web-panacea.git

Commands

Run Benchmark

$ npm run start

or

$ npm start

Then you should follow the instructions of CLI.

How to add a new test subject?

It is recommended to add the source code of the test subject to the appropriate language in the folder in the subjects folder. For example, you add the source code of some Java-framework: subjects/java/some-framework <- sources here

Then you have to add a test subject to settings/subjects.json and create the new object:

{
  // previous subjects
  "Some Java Framework": {
    "dir": "java/some-framework", // Source folder
    "build": "mvn clean package", // Command to start the build process
    "run": { // Command to start the built project
      "command": "java",
      "args": ["-jar", "target/some-framework.jar"] // Keys
    },
    "supports": [ // Implemented endpoints for testing (benchmark-types)
      "get-tiny-json-entity",
      "get-plain-text"
    ]
  }
}

Important The subject should print a "START" signal at startup.

How to add a new benchmark type?

You can add a new type in the benchmark types file:

{
  // previous types
  "post-some-json-entity": {
    "url": "http://localhost:8080/api/post-some-json-entity",
    "method": "POST",
    "headers": { "Content-Type": "application/json" },
    "body": "{\"some\":\"entity\"}"
  }
}

In-Search-of-a-Web-Panacea uses autocannon, so additional fields can be found in its documentation.

Author

Egor Repnikov - GitHub

License

This project is licensed under the MIT License - see the LICENSE file for details