Pocket load tester

Scale curl requests, cousin of ab, siege, hey.

Demo

Install

go install github.com/vearutop/plt@latest
$(go env GOPATH)/bin/plt --help

Or download binary from releases.

Linux AMD64

wget -q https://github.com/vearutop/plt/releases/latest/download/linux_amd64.tar.gz && tar xf linux_amd64.tar.gz && rm linux_amd64.tar.gz
./plt -version

Usage

usage: plt [<flags>] <command> [<args> ...]

Pocket load tester pushes to the limit

Flags:
  --help            Show context-sensitive help (also try --help-long and --help-man).
  --number=1000     Number of requests to run, 0 is infinite.
  --concurrency=50  Number of requests to run concurrently.
  --rate-limit=0    Rate limit, in requests per second, 0 disables limit (default).
  --duration=1m     Max duration of load testing, 0 is infinite.
  --slow=1s         Min duration of slow response.
  --live-ui         Show live ui with statistics.

Commands:
  help [<command>...]
    Show help.

  curl [<flags>] [<url>]
    Repetitive HTTP transfer

You can "copy as cURL" in your browser and then prepend that with plt to throw 1000 of such requests.

For even better performance you can use plt curl --fast that will employ awesome fasthttp as transport. This mode lacks detailed breakdown of latencies, but can push request rate to the limit.

Use --http2 for HTTP/2 or --http3 for HTTP/3 (go1.19 or later).

If the server is wrapped with Envoy proxy, upstream latency distribution will be collected from the values of X-Envoy-Upstream-Service-Time response header.

In --live-ui mode you can control concurrency and rate limits with arrow keys.

Sending different requests

Sending same request over and over again is not very useful in some cases. It may hit caches or be blocked by the server. However, describing dynamic requests with predefined parameters is very difficult in general that's why plt does not attempt to provide such functionality.

Instead, you can extend plt with your own custom request preparer written in Go and having full control and flexibility with familiar API. Please check an example of customization with dynamic requests.

Example

plt --live-ui --duration=20s --rate-limit=60 curl -X GET "https://demo.phperf.ml/profile"

plt --live-ui --duration=20s --rate-limit=60 curl -X GET "https://demo.phperf.ml/profile"
Host resolved: 130.61.99.204

Requests per second: 59.96
Total requests: 1201
Time spent: 20.029s

Request latency percentiles:
99%: 101.73ms
95%: 64.88ms
90%: 44.57ms
50%: 28.09ms

Request latency distribution in ms:
[   min    max]  cnt total% (1201 events)
[ 19.03  19.03]    1  0.08%
[ 22.52  23.76]   25  2.08% ..
[ 23.78  25.26]  111  9.24% .........
[ 25.26  26.58]  200 16.65% ................
[ 26.58  29.61]  449 37.39% .....................................
[ 29.61  39.94]  287 23.90% .......................
[ 40.54  40.54]    1  0.08%
[ 41.00  62.45]   60  5.00% ....
[ 62.77 110.91]   57  4.75% ....
[126.64 260.53]   10  0.83%

Requests with latency more than 1s: 0

Bytes read 2.3MB / 23.5MB/s
Bytes written 105.4KB / 57.6KB/s

DNS latency distribution in ms:
[ min  max] cnt total% (16 events)
[0.44 0.44]  1  6.25% ......
[0.49 0.49]  1  6.25% ......
[0.50 0.50]  1  6.25% ......
[0.56 0.56]  1  6.25% ......
[0.61 0.61]  2 12.50% ............
[0.65 0.70]  3 18.75% ..................
[0.77 0.77]  1  6.25% ......
[0.88 0.88]  1  6.25% ......
[1.00 1.03]  3 18.75% ..................
[1.28 1.42]  2 12.50% ............

TLS handshake latency distribution in ms:
[   min    max] cnt total% (16 events)
[ 64.95  64.95]  1  6.25% ......
[ 65.14  65.14]  1  6.25% ......
[ 66.98  66.98]  1  6.25% ......
[ 70.94  71.81]  3 18.75% ..................
[ 74.10  76.13]  2 12.50% ............
[ 80.20  82.91]  2 12.50% ............
[ 90.86  90.86]  1  6.25% ......
[127.14 127.72]  2 12.50% ............
[157.99 161.84]  2 12.50% ............
[179.98 179.98]  1  6.25% ......

Time to first resp byte (TTFB) distribution in ms:
[   min    max]  cnt total% (1201 events)
[ 18.97  18.97]    1  0.08%
[ 22.49  23.77]   27  2.25% ..
[ 23.78  24.96]   81  6.74% ......
[ 24.96  26.56]  236 19.65% ...................
[ 26.57  29.52]  441 36.72% ....................................
[ 29.54  29.54]    1  0.08%
[ 29.56  40.90]  288 23.98% .......................
[ 41.16  62.68]   60  5.00% ....
[ 63.02 110.87]   56  4.66% ....
[126.59 260.27]   10  0.83%

Connection latency distribution in ms:
[  min   max] cnt total% (16 events)
[28.04 28.04]  1  6.25% ......
[28.13 28.13]  1  6.25% ......
[29.68 29.68]  1  6.25% ......
[30.01 30.41]  2 12.50% ............
[31.04 31.12]  2 12.50% ............
[32.22 32.44]  2 12.50% ............
[33.37 34.07]  3 18.75% ..................
[36.04 36.04]  1  6.25% ......
[40.09 40.09]  1  6.25% ......
[43.50 45.20]  2 12.50% ............

Responses by status code
[200] 1201

Response samples (first by status code):
[200]
Connection: keep-alive
Content-Length: 1733
Content-Type: application/json; charset=utf-8
Date: Thu, 08 Apr 2021 15:25:54 GMT
Server: nginx/1.14.0 (Ubuntu)

{"recent":[{"addr":{"id":"606f1f90dcdc3"},"edges":3715,"wt":"190.55ms","cpu":"179.03ms","io":"11.52ms","peakMem":"3.52M"},{"addr":{"id":"606f1faf1c5c3"},"edges":3715,"wt":"270.14ms","cpu":"185.62ms","io":"84.52ms","peakMem":"3.52M"},{"addr":{"id":"606f1fc261e9d"},"edges":575,"wt":"3.4s","cpu":"3.39s","io":"9.71ms","peakMem":"28.03M"},{"addr":{"id":"606f1fcd6f694"},"edges":3715,"wt":"153.58ms","cpu":"143.68ms","io":"9.9ms","peakMem":"3.52M"},{"addr":{"id":"606f1feba911d"},"edges":3715,"wt":"202.18ms","cpu":"191.82ms","io":"10.36ms","peakMem":"3.52M"},{"addr":{"id":"606f20052f7c1"},"edges":471,"wt":"679.08ms","cpu":"669.01ms","io":"10.07ms","peakMem":"6.18M"},{"addr":{"id":"606f2009de4cd"},"edges":3715,"wt":"175.34ms","cpu":"163.14ms","io":"12.2ms","peakMem":"3.52M"},{"addr":{"id":"606f2028185dd"},"edges":3715,"wt":"677.03ms","cpu":"320.84ms","io":"356.19ms","peakMem":"3.52M"},{"addr":{"id":"606f2046c16a2"},"edges":3715,"wt":"313.28ms","cpu":"292.38ms","io":"20.9ms","peakMem":"3.52M"},{"...

S3 latency

usage: plt s3 --bucket=BUCKET --key=KEY [<flags>]

S3 transfer

Flags:
  --help                         Show context-sensitive help (also try --help-long and --help-man).
  --version                      Show application version.
  --number=1000                  Number of requests to run, 0 is infinite.
  --concurrency=50               Number of requests to run concurrently.
  --rate-limit=0                 Rate limit, in requests per second, 0 disables limit (default).
  --duration=1m                  Max duration of load testing, 0 is infinite.
  --slow=1s                      Min duration of slow response.
  --live-ui                      Show live ui with statistics.
  --access-key=ACCESS-KEY        Access key/id (env AWS_ACCESS_KEY).
  --secret-key=SECRET-KEY        Secret key (env AWS_SECRET_KEY).
  --session-token=SESSION-TOKEN  Session token (env AWS_SESSION_TOKEN).
  --region="eu-central-1"        Region.
  --url=URL                      Optional S3 URL (if not AWS).
  --bucket=BUCKET                Bucket name.
  --key=KEY                      Entry key.
  --path-style                   To use path-style addressing, i.e., `http://s3.amazonaws.com/BUCKET/KEY`.
  --save=SAVE                    Path to local file to save the entry.
  --upload=UPLOAD                Path to local file to upload to S3, enables upload load testing.

./plt --number 1000 --rate-limit 300 s3 --access-key=<redacted> --secret-key=<redacted> --url=ceph.example.com:7480 --bucket=foo --key=bar --path-style

Requests per second: 299.34
Successful requests: 1000
Time spent: 3.341s

Request latency percentiles:
99%: 46.36ms
95%: 15.99ms
90%: 9.35ms
50%: 7.01ms

Request latency distribution in ms:
[  min   max]  cnt total% (1000 events)
[ 6.44  6.44]    1  0.10%
[ 6.49  6.55]    9  0.90%
[ 6.55  6.66]   43  4.30% ....
[ 6.66  7.00]  445 44.50% ............................................
[ 7.00  7.89]  346 34.60% ..................................
[ 7.93  7.93]    1  0.10%
[ 7.93 10.39]   70  7.00% .......
[10.54 13.57]   23  2.30% ..
[13.65 28.34]   30  3.00% ...
[30.16 52.90]   32  3.20% ...

Requests with latency more than 1s: 0

Read: total 2600.89 MB, avg 2.60 MB, 778.49 MB/s

Beautify the output

Here's an example output of the tool:

❯ plt --live-ui --concurrency=5 curl http://localhost:8000/login
Host resolved: ::1,127.0.0.1
Requests per second: 33.97
Total requests: 1000
Request latency distribution in ms:
[   min    max]  cnt total% (1000 events)
[ 53.05  53.05]    1  0.10%
[ 53.44  53.63]    2  0.20%
[ 57.30  57.30]    1  0.10%
[ 61.55  61.55]    1  0.10%
[ 64.40  72.33]   10  1.00% .
[ 74.08  86.93]   24  2.40% ..
[ 87.01 120.37]  201 20.10% ....................
[120.48 155.74]  414 41.40% .........................................
[155.75 188.30]  267 26.70% ..........................
[188.71 304.20]   79  7.90% .......

Request latency percentiles:
99%: 240.994161ms
95%: 202.252658ms
90%: 184.659317ms
50%: 144.925221ms

Requests with latency more than 1s: 0

DNS latency distribution in ms:
[ min  max] cnt total% (12 events)
[0.53 0.53]  1  8.33% ........
[0.57 0.57]  1  8.33% ........
[0.58 0.59]  2 16.67% ................
[0.60 0.60]  1  8.33% ........
[0.62 0.62]  2 16.67% ................
[0.63 0.63]  1  8.33% ........
[0.65 0.65]  1  8.33% ........
[0.74 0.74]  1  8.33% ........
[1.06 1.06]  1  8.33% ........
[1.11 1.11]  1  8.33% ........

Connection latency distribution in ms:
[ min  max] cnt total% (12 events)
[0.21 0.21]  1  8.33% ........
[0.22 0.22]  1  8.33% ........
[0.23 0.23]  1  8.33% ........
[0.23 0.23]  2 16.67% ................
[0.25 0.25]  1  8.33% ........
[0.31 0.31]  1  8.33% ........
[0.36 0.36]  1  8.33% ........
[0.42 0.43]  2 16.67% ................
[0.48 0.48]  1  8.33% ........
[1.01 1.01]  1  8.33% ........

Responses by status code
[200] 1000

Bytes read 5870717
Bytes written 85000
[200]
<!doctype html>
<html lang="en">
<head>
    <meta charset="utf-8">
    <meta name="viewport" content="width=device-width, initial-scale=1">

    <!-- CSRF Token -->
    <meta name="csrf-token" content="gpUQ1qVPOT165dVG1rqkUZvfpLBIDiU0KLIjT9R4">

    <title>Laravel</title>

    <!-- Styles -->
    <link href="/css/app.css" rel="stylesheet">
</head>
<body class="bg-gray-200 h-screen antialiased leading-none">
    <div id="app">
            <div class="container mx-auto">
        <main role="main" class="px-4 md:px-8">
                <div class="flex flex-wrap items-center justify-center min-h-screen">
        <div class="w-full md:w-1/3">
            <h2 class="font-extrabold leading-9 mb-6 text-3xl text-center text-gray-700">
                Login to your account
            </h2>
            <div class="bg-white break-words flex flex-col rounded-md shadow-md">
                <form class="w-full p-6" method="POST" action="http://localhost:8000/login">
                    <input type="...

Reading the above is confusing and requires carefully checking the output. It doesn't really have a hierarchy, and shows information that is not necessarily required. Let's have a look at ab output for the same site:

❯ ab localhost:8000/login
This is ApacheBench, Version 2.3 <$Revision: 1874286 $>
Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
Licensed to The Apache Software Foundation, http://www.apache.org/

Benchmarking localhost (be patient).....done


Server Software:        nginx/1.17.5
Server Hostname:        localhost
Server Port:            8000

Document Path:          /login
Document Length:        4187 bytes

Concurrency Level:      1
Time taken for tests:   0.461 seconds
Complete requests:      1
Failed requests:        0
Total transferred:      5858 bytes
HTML transferred:       4187 bytes
Requests per second:    2.17 [#/sec] (mean)
Time per request:       460.860 [ms] (mean)
Time per request:       460.860 [ms] (mean, across all concurrent requests)
Transfer rate:          12.41 [Kbytes/sec] received

Connection Times (ms)
              min  mean[+/-sd] median   max
Connect:        0    0   0.0      0       0
Processing:   461  461   0.0    461     461
Waiting:      459  459   0.0    459     459
Total:        461  461   0.0    461     461

As you can tell, the output of ab is much easier to process, although it doesn't provide as much details as plt does.

Proposed Output

Here's a quick sketch of how the output might be changed:

❯ plt --live-ui --concurrency=5 curl http://localhost:8000/login

--------------------------------------------

Successfully completed in 17.89 seconds.

Latency Distribution
+---------+---------+-------+------------+
| Minimum | Maximum | Count | Percentage |
+---------+---------+-------+------------+
|   53.05 |   53.05 |     1 | 0.10       |
|   53.44 |   53.63 |     2 | 0.20       |
|   57.30 |   57.30 |     1 | 0.10       |
|   61.55 |   61.55 |     1 | 0.10       |
|   64.40 |   72.33 |    10 | 1.00%      |
|   74.08 |   86.93 |    24 | 2.40%      |
|   87.01 |  120.37 |   201 | 20.10%     |
|  120.48 |  155.74 |   414 | 41.40%     |
|  155.75 |  188.30 |   267 | 26.70%     |
|  188.71 |  304.20 |    79 | 7.90%      |
+---------+---------+-------+------------+

Latency Percentiles
+-----+--------------+
| Pp  |   Latency    |
+-----+--------------+
| 99% | 240.994161ms |
| 95% | 202.252658ms |
| 90% | 184.659317ms |
| 50% | 144.925221ms |
+-----+--------------+

Status Code Distribution
+-------------+-------+
| Status Code | Count |
+-------------+-------+
|         200 |  1000 |
+-------------+-------+

DNS Resolution Latency
+---------+---------+-------+------------+
| Min(ms) | Max(ms) | Count | Percentage |
+---------+---------+-------+------------+
|    0.53 |    0.53 |     1 | 8.33%      |
|    0.57 |    0.57 |     1 | 8.33%      |
|    0.58 |    0.59 |     2 | 16.67%     |
|    0.60 |    0.60 |     1 | 8.33%      |
|    0.62 |    0.62 |     2 | 16.67%     |
|    0.63 |    0.63 |     1 | 8.33%      |
|    0.65 |    0.65 |     1 | 8.33%      |
|    0.74 |    0.74 |     1 | 8.33%      |
|    1.06 |    1.06 |     1 | 8.33%      |
|    1.11 |    1.11 |     1 | 8.33%      |
+---------+---------+-------+------------+

Connection Latency
+---------+---------+-------+------------+
| Min(ms) | Max(ms) | Count | Percentage |
+---------+---------+-------+------------+
|    0.53 |    0.53 |     1 | 8.33%      |
|    0.57 |    0.57 |     1 | 8.33%      |
|    0.58 |    0.59 |     2 | 16.67%     |
|    0.60 |    0.60 |     1 | 8.33%      |
|    0.62 |    0.62 |     2 | 16.67%     |
|    0.63 |    0.63 |     1 | 8.33%      |
|    0.65 |    0.65 |     1 | 8.33%      |
|    0.74 |    0.74 |     1 | 8.33%      |
|    1.06 |    1.06 |     1 | 8.33%      |
|    1.11 |    1.11 |     1 | 8.33%      |
+---------+---------+-------+------------+

----------------------------------------------

Total Requests:        1000          
Failed Requests:       0             
Requests per second:   33.97         
p99 Latency:           240.994161ms  
p50 Latency:           240.994161ms

Combined with font weights and colors, this might make the output much easier to process.

vearutop / plt Goto Github PK

plt's Introduction

Pocket load tester

Demo

Install

Linux AMD64

Usage

Sending different requests

Example

S3 latency

plt's People

Contributors

Stargazers

Watchers

Forkers

plt's Issues

Proposed Output

Recommend Projects

Recommend Topics

Recommend Org