/ttfb.sh

Measures time-to-first-byte in seconds, for single or multiple URLs. Can calculate fastest, slowest & median TTFB values, and optionally log all response headers. Uses curl and the calculation `%{time_starttransfer} - %{time_appconnect}` which doesn't include any connection overhead, to better approximate devtool’s TTFB figure.

Primary LanguageShellMIT LicenseMIT

Introduction

Measures time-to-first-byte for single or multiple URLs. Can show you quickest, slowest & median TTFB values plus optionally log all response headers.

Sample run of ttfb.sh

Usage: ttfb [options] url [url...]
	-d debug
	-l <log file> (infers -d) log response headers. Defaults to ./curl.log
	-n <number> of times to test time to first byte
	-v verbose output. Show response breakdown (DNS lookup, TLS handshake etc)

Implicitly follows a redirection chain using curl's -L option.

Can log all response headers (the default log file is ./curl.log) by calling with -d.

Override the default log file by specifying -l /some/file.

Get quickest, slowest and median TTFB values by specifying the number of times to call a URL; use -n2 for 2 tests, -n5 for 5 and so on.

Uses the calculation %{time_starttransfer¹} - %{time_appconnect²} which doesn't include any connection overhead, to better approximate devtool’s TTFB figure.

¹ time_starttransfer

The time, in seconds, it took from the start until the first byte was just about to be transferred. This includes time_pretransfer and also the time the server needed to calculate the result.

² time_appconnect

The time, in seconds, it took from the start until the SSL/SSH/etc connect/handshake to the remote host was completed.

Genesis

Based on a gist by https://github.com/sandeepraju

Modified by jay@gooby.org, @jaygooby

Usage

Usage: ttfb [options] url [url...]
	-d debug
	-l <log file> (infers -d) log response headers. Defaults to ./curl.log
	-n <number> of times to test time to first byte
	-v verbose output. Show response breakdown (DNS lookup, TLS handshake etc)

Examples

Basic usage:

$ ttfb example.com
.227436

Basic usage with verbose response breakdown:

$ ttfb -v https://example.com
DNS lookup: 0.005152 TLS handshake: 0.000000 TTFB including connection: 0.200831 TTFB: .200831 Total time: 0.201132

Test multiple times:

$ ttfb -n 5 example.com/example/url
.....
fastest .177263 slowest .214302 median .179957

Test multiple URLs:

$ ttfb bbc.co.uk news.bbc.co.uk
bbc.co.uk        .049985
news.bbc.co.uk   .054122

Test multiple URLs, multiple times:

$ ttfb -n 5 bbc.co.uk news.bbc.co.uk
.....
.....
bbc.co.uk       fastest .030936 slowest .057755 median .034663
news.bbc.co.uk  fastest .031413 slowest .182791 median .035001

Verbose response breakdown when multiple tests specified:

$ ttfb -v -n 5 bbc.co.uk
DNS lookup: 0.005335 TLS handshake: 0.102314 TTFB including connection: 0.148328 TTFB: .046014 Total time: 0.646115
DNS lookup: 0.005322 TLS handshake: 0.102609 TTFB including connection: 0.150693 TTFB: .048084 Total time: 0.644611
DNS lookup: 0.004277 TLS handshake: 0.102066 TTFB including connection: 0.172199 TTFB: .070133 Total time: 1.196256
DNS lookup: 0.004444 TLS handshake: 0.107375 TTFB including connection: 0.160771 TTFB: .053396 Total time: 0.637290
DNS lookup: 0.005352 TLS handshake: 0.118882 TTFB including connection: 0.168772 TTFB: .049890 Total time: 0.653761

fastest .046014 slowest .070133 median .049890

Log all the response headers for multiple tests to multiple URLs:

ttfb -d -n 2 bbc.co.uk https://www.bbc.co.uk/weather
..
..
bbc.co.uk                      fastest .027550 slowest .055215 median .041382
https://www.bbc.co.uk/weather  fastest .101020 slowest .297923 median .199471

$ ls *.log
bbc_co_uk-curl.log                     https___www_bbc_co_uk_weather-curl.log

$ cat https___www_bbc_co_uk_weather-curl.log
HTTP/2 200
server: openresty
x-cache-action: MISS
vary: Accept-Encoding,X-BBC-Edge-Cache,X-BBC-Edge-Scheme,X-CDN
x-cache-age: 0
cache-control: private, stale-while-revalidate=10, max-age=0, must-revalidate
content-type: text/html;charset=utf-8
x-mrid: w1
date: Thu, 11 Apr 2019 17:08:07 GMT
x-xss-protection: 1; mode=block
x-content-type-options: nosniff
x-lb-nocache: true
x-msig: 24e37f81323984e4e45b8048f9e3c94a
x-frame-options: SAMEORIGIN
content-length: 1077454

HTTP/2 200
server: openresty
x-cache-action: MISS
vary: Accept-Encoding,X-BBC-Edge-Cache,X-BBC-Edge-Scheme,X-CDN
x-cache-age: 0
cache-control: private, stale-while-revalidate=10, max-age=0, must-revalidate
content-type: text/html;charset=utf-8
x-mrid: w1
date: Thu, 11 Apr 2019 17:08:08 GMT
x-xss-protection: 1; mode=block
x-content-type-options: nosniff
x-lb-nocache: true
x-msig: 24e37f81323984e4e45b8048f9e3c94a
x-frame-options: SAMEORIGIN
content-length: 1077454

More detail on time-to-first-byte

See https://blog.cloudflare.com/a-question-of-timing/ and https://curl.haxx.se/docs/manpage.html for an explanation of how the curl variables relate to the various stages of the transfer.

Diagram showing what each of the curl variable timings refer to against a typical HTTP over TLS 1.2 connection

To get a better approximation of devtool's TTFB, we consider the time without the connection overhead: %{time_starttransfer} - %{time_appconnect}

Uses a dirty eval to do the ttfb arithmetic. Depends on bc and column commands.

TODO

  • Show progress when more than one request (-n 2 etc) option is set

  • Sort output by fastest TTFB when multiple URLs are supplied

  • Colour code the TTFB: figure in the standard response, according to the speed of the response.