tyk benchmark

readme.md

Go Plugin vs JSVM Middleware Benchmark

Date: 2026-02-20 | Tool: Apache Bench | Requests: 10,000 | Concurrency: 10

Tyk version: commit c0413c845222d413ae82d2ce68f805a8839efb23 Version ~ v5.12.0

Commands

ab -n 10000 -c 10 http://localhost:8091/gopl/hello/
ab -n 10000 -c 10 http://localhost:8091/jsvm/hello/

Results

Metric	Go Plugin	JSVM	Difference
Requests/sec	14,247	1,130	Go is ~12.6x faster
Mean latency	0.70 ms	8.85 ms	Go is ~12.6x lower
Wall time	0.70 s	8.85 s
CPU time consumed	2.32 s	21.51 s	Go uses ~9.3x less CPU
CPU utilization	~331%	~243%	Go saturates more cores

Summary

The Go plugin middleware is roughly 13x faster than JSVM in throughput and uses ~9x less total CPU time for the same workload. JSVM adds ~8ms overhead per request from the JavaScript VM execution context. Go's higher CPU utilization (331% vs 243%) shows it effectively parallelizes across cores, while JSVM is bottlenecked by VM execution.

goplg-hello.json

{
    "name": "Go Plugin Hello",
    "api_id": "gopl-hello",
    "org_id": "default",
    "active": true,
    "use_keyless": true,
    "definition": {
        "location": "header",
        "key": "x-api-version"
    },
    "version_data": {
        "not_versioned": true,
        "versions": {
            "Default": {
                "name": "Default",
                "use_extended_paths": true
            }
        }
    },
    "proxy": {
        "listen_path": "/gopl/",
        "target_url": "http://localhost:8222",
        "strip_listen_path": true
    },
    "custom_middleware": {
        "driver": "goplugin",
        "pre": [
            {
                "name": "AddHeader",
                "path": "middleware/go-plugin/plugin.so"
            }
        ]
    }
}

jsvm-hello.json

{
    "name": "JSVM Hello",
    "api_id": "jsvm-hello",
    "org_id": "default",
    "active": true,
    "use_keyless": true,
    "definition": {
        "location": "header",
        "key": "x-api-version"
    },
    "version_data": {
        "not_versioned": true,
        "versions": {
            "Default": {
                "name": "Default",
                "use_extended_paths": true
            }
        }
    },
    "proxy": {
        "listen_path": "/jsvm/",
        "target_url": "http://localhost:8222",
        "strip_listen_path": true
    },
    "custom_middleware": {
        "driver": "otto",
        "pre": [
            {
                "name": "jsvmHelloPreHook",
                "path": "middleware/jsvm_hello.js"
            }
        ]
    }
}

jsvm_hello.js

var jsvmHelloPreHook = new TykJS.TykMiddleware.NewMiddleware({});

jsvmHelloPreHook.NewProcessRequest(function(request, session) {
    request.SetHeaders["X-Custom-JSVM"] = "hello-from-jsvm";
    return jsvmHelloPreHook.ReturnData(request, {});
});

log("jsvm_hello.js middleware loaded");

local-mock-server.go

// +build ignore

package main

import (
	"encoding/json"
	"fmt"
	"net/http"
	"os"
)

func main() {
	http.HandleFunc("/", func(w http.ResponseWriter, r *http.Request) {
		headers := make(map[string]string)
		for k, v := range r.Header {
			if len(v) > 0 {
				headers[k] = v[0]
			}
		}
		resp := map[string]interface{}{
			"message": "hello from mock upstream",
			"path":    r.URL.Path,
			"method":  r.Method,
			"headers": headers,
		}
		w.Header().Set("Content-Type", "application/json")
		json.NewEncoder(w).Encode(resp)
	})

	port := "8222"
	if p := os.Getenv("MOCK_PORT"); p != "" {
		port = p
	}
	fmt.Printf("Mock server listening on :%s\n", port)
	http.ListenAndServe(":"+port, nil)
}

main.go

package main

import (
	"net/http"
)

// AddHeader adds a custom header to every request.
func AddHeader(rw http.ResponseWriter, r *http.Request) {
	r.Header.Set("X-Custom-GoPlugin", "hello-from-go-plugin")
}

func main() {}