PhantomJS-based modular web performance metrics collector. And why phantomas? Well, because :)
npm install --global phantomas
This will install the latest version of PhantomJS and add a symlink called
phantomas
(pointing to./bin/phantomas.js
) to your system'sPATH
- Modular approach - each metric is generated by a separate "module"
- phantomas "core" acts as an events emitter that each module can hook into
- in-depth metrics such as: number of events bound via jQuery, calls to
window.write
or complex and duplicated CSS selectors (via analyze-css) - JSON and CSV as available output formats for easy integration with automated reporting / monitoring tools
- easy integration with Continuous Integration tools via TAP format and assertions handling
- easy integration with other nodejs projects via CommonJS module (see API docs)
- metrics can be emitted from JavaScript code of the page phantomas is run against (thanks to helper functions available in window.__phantomas)
- macbre
- jmervine
- jmosney
- umaar
- sjhcockrell
- cphoover
- LaurentGoderre
- kennydee
- iNem0o
- stefanjudis
- vgangan
- All the contributors
phantomas comes as both command line tool and CommonJS module (see API docs) that you can use in your nodejs projects.
phantomas https://github.com/macbre/phantomas --verbose
You can measure the performance of your site without requests to 3rd party domains (but allowing CDN that serves your static assets):
phantomas https://github.com/macbre/phantomas --verbose --no-externals --allow-domain .fastly.net
--reporter=[json|csv|tap|plain|statsd|elasticsearch]
results reporter aka format (plain
is the default one)--timeout=[seconds]
timeout for phantomas run (defaults to 15 seconds)--viewport=[width]x[height]
phantomJS viewport dimensions (1280x1024 is the default)--verbose
writes debug messages to the console--silent
don't write anything to the console--progress
shows page loading progress bar (disables verbose mode)--log=[log file]
log to a given file--modules=[moduleOne],[moduleTwo]
run only selected modules--skip-modules=[moduleOne],[moduleTwo]
skip selected modules--user-agent='Custom user agent'
provide a custom user agent (will default to something similar tophantomas/0.6.0 (PhantomJS/1.9.0; linux 64bit)
)--config=[JSON config file]
uses JSON-formatted config file to set parameters--cookie='bar=foo;domain=url'
document.cookie formatted string for setting a single cookie--cookies-file=[JAR file]
specifies the file name to store the persistent Cookies--no-externals
block requests to 3rd party domains--allow-domain=[domain],[domain]
allow requests to given domain(s) - aka whitelist--block-domain=[domain],[domain]
disallow requests to given domain(s) - aka blacklist--disable-js
disable JavaScript on the page that will be loaded--analyze-css
emit in-depth CSS metrics experimental--film-strip
register film strip when page is loading experimental--film-strip-dir=[dir path]
folder path to output film strip (default is./filmstrip
directory) experimental--assert-[metric-name]=value
assert that given metric should be less or equal the value--screenshot=[file name]
render fully loaded page to a given file--wait-for-selector=[CSS selector]
wait for an element matching given CSS selector before generating a report, timeout setting still applies (e.g.--wait-for-selector "body.loaded"
)--post-load-delay=[seconds]
wait X seconds before generating a report, timeout setting still applies--ignore-ssl-errors
ignores SSL errors, such as expired or self-signed certificate errors--proxy=[host:port]
specifies the proxy server to use--proxy-auth=[username:password]
specifies the authentication information for the proxy--proxy-type=[http|socks5|none]
specifies the type of the proxy server (default is http)--phone
force viewport and user agent of a mobile phone--tablet
force viewport and user agent of a tablet
Simply provide --runs
option:
phantomas https://github.com/macbre/phantomas --verbose --runs 5
Only plain
(the default one) and json
reporters are currently supported in multiple runs mode.
Current number of metrics: 102
Units:
- ms for time
- bytes for size
Due to PhantomJS issue #10156 body size related metrics are not reliable
- requests: total number of HTTP requests made
- gzipRequests: number of gzipped HTTP responses
- postRequests: number of POST requests
- httpsRequests: number of HTTPS requests
- notFound: number of HTTP 404 responses
- timeToFirstByte: time it took to receive the first byte of the first response (that was not a redirect)
- timeToLastByte: time it took to receive the last byte of the first response (that was not a redirect)
- bodySize: size of the content of all responses
- contentLength: size of the content of all responses (based on
Content-Length
header) - httpTrafficCompleted: time it took to receive the last byte of the last HTTP response
- ajaxRequests: number of AJAX requests
Due to PhantomJS issue #10156 body size related metrics are not reliable
- htmlCount: number of HTML responses
- htmlSize: size of HTML responses
- cssCount: number of CSS responses
- cssSize: size of CSS responses
- jsCount: number of JS responses
- jsSize: size of JS responses
- jsonCount: number of JSON responses
- jsonSize: size of JSON responses
- imageCount: number of image responses
- imageSize: size of image responses
- webfontCount: number of web font responses
- webfontSize: size of web font responses
- base64Count: number of base64 encoded "responses" (no HTTP request was actually made)
- base64Size: size of base64 encoded "responses"
- otherCount: number of other responses
- otherSize: size of other responses
Metrics are calculated based on
X-Cache
header added by Varnish / Squid servers
- cacheHits: number of cache hits
- cacheMisses: number of cache misses
- cachePasses: number of cache passes
- headersCount: number of requests and responses headers
- headersSentCount: number of headers sent in requests
- headersRecvCount: number of headers received in responses
- headersSize: size of all headers
- headersSentSize: size of sent headers
- headersRecvSize: size of received headers
- domains: number of domains used to fetch the page
- maxRequestsPerDomain: maximum number of requests fetched from a single domain
- medianRequestsPerDomain: median of requests fetched from each domain
- cookiesSent: length of cookies sent in HTTP requests
- cookiesRecv: length of cookies received in HTTP responses
- domainsWithCookies: number of domains with cookies set
- documentCookiesLength: length of
document.cookie
- documentCookiesCount: number of cookies in
document.cookie
Metrics listed below are generated after the full page load
- globalVariables: number of JS globals variables
- globalVariablesFalsy: number of JS global variables that cast to false
- bodyHTMLSize: the size of body tag content (
document.body.innerHTML.length
) - commentsSize: the size of HTML comments on the page
- hiddenContentSize: the size of content of hidden elements on the page (with CSS
display: none
) - whiteSpacesSize: the size of text nodes with whitespaces only
- DOMelementsCount: total number of HTML element nodes
- DOMelementMaxDepth: maximum level on nesting of HTML element node
- iframesCount: number of iframe nodes
- nodesWithInlineCSS: number of nodes with inline CSS styling (with
style
attribute) - imagesWithoutDimensions: number of
<img>
nodes without bothwidth
andheight
attribute
- DOMqueries: the sum of all four metrics below
- DOMqueriesById: number of
document.getElementById
calls - DOMqueriesByClassName: number of
document.getElementsByClassName
calls - DOMqueriesByTagName: number of
document.getElementsByTagName
calls - DOMqueriesByQuerySelectorAll: number of
document.querySelectorAll
calls - DOMinserts: number of DOM nodes inserts
- DOMqueriesDuplicated: number of duplicated DOM queries
- eventsBound: number of
EventTarget.addEventListener
calls
Times below are relative to
responseEnd
entry in NavigationTiming (represented bytimeToLastByte
metric). See NavigationTiming spec for more information.
- onDOMReadyTime: time it took to fire onDOMready event
- onDOMReadyTimeEnd: time it took to finish processing onDOMready event experimental
- windowOnLoadTime: time it took to fire window.load event
- windowOnLoadTimeEnd: time it took to finish processing window.load event experimental
- timeBackend: time to the first byte compared to the total loading time (in %)
- timeFrontend: time to window on load compared to the total loading time (in %)
Time is total duration, from the start of the request to the receipt of the final byte in the response. Latency is the time to load the first byte in the response. https://developers.google.com/chrome-developer-tools/docs/network
Includes
HTTP 200
responses only
- smallestResponse: the size of the smallest response
- biggestResponse: the size of the biggest response
- fastestResponse: the time to the last byte of the fastest response
- slowestResponse: the time to the last byte of the slowest response
- smallestLatency: the time to the first byte of the fastest response
- biggestLatency: the time to the first byte of the slowest response
- medianResponse: median value of time to the last byte for all responses
- medianLatency: median value of time to the first byte for all responses
- localStorageEntries: number of entries in local storage
Requires jQuery 1.8.0+
- jQueryVersion: version of jQuery framework (if loaded)
- jQueryOnDOMReadyFunctions: number of functions bound to onDOMReady event
- jQuerySizzleCalls: number of calls to Sizzle (including those that will be resolved using
querySelectorAll
)
- assetsNotGzipped: static assets that were not gzipped
- assetsWithQueryString: static assets requested with query string (e.g. ?foo) in URL
- smallImages: images smaller than 2 kB that can be base64 encoded
- multipleRequests: number of static assets that are requested more than once
- cachingNotSpecified: responses with no caching header sent (either
Cache-Control
orExpires
) - cachingTooShort: responses with too short (less than a week) caching time
- cachingDisabled: responses with caching disabled (
max-age=0
) - oldCachingHeaders: responses with old, HTTP 1.0 caching headers (
Expires
andPragma
)
- timeToFirstCss: time it took to receive the last byte of the first CSS
- timeToFirstJs: time it took to receive the last byte of the first JS
- timeToFirstImage: time it took to receive the last byte of the first image
- redirects: number of HTTP redirects (either 301, 302 or 303)
- redirectsTime: time it took to send and receive redirects
- documentWriteCalls: number of calls to either
document.write
ordocument.writeln
- evalCalls: number of calls to
eval
(either direct or viasetTimeout
/setInterval
)
Error message and backtrace will be emitted as offenders
- jsErrors: number of JavaScript errors
- windowAlerts: number of calls to
alert
- windowConfirms: number of calls to
confirm
- windowPrompts: number of calls to
prompt
- consoleMessages: number of calls to
console.*
functions
Analyzes bits of data pertaining to the main request only
- statusCodesTrail: comma-separated list of HTTP status codes that main request followed through (could contain a single element if the main request is a terminal one)
The following metrics are emitted only when certain options are passed to phantomas
- blockedRequests: number of requests blocked due to domain filtering (emitted only when in
--no-externals
/--block-domain
mode)
This is an experimental feature. Use
--analyze-css
option to enable it.
Take a look at analyze-css README for the full list of metrics.
phantomas provides a number of reporters that can format the run results and send them to various tools. Use --reporter
(or -R
shortcut) option to use one.
Results can be emitted as TAP, CSV and JSON. plain
format is most useful for human beings :)
Metrics from phantomas run can be sent directly to StatsD and then graphed using graphite, graphene or any other tool of your choice. For instance:
$ phantomas http://app.net/start -R statsd --statsd-host stats.app.net --statsd-port 8125 --statsd-prefix 'myApp.mainPage.'
will sent metrics to StatsD running on stats.app.net:8125
and prefix them with 'myApp.mainPage'.
Metrics from phantomas run can be outputted directly in ElasticSearch :
--elasticsearch-host=[ip]
ElasticSearch instance ip (default : 127.0.0.1)--elasticsearch-port=[port]
ElasticSearch instance port (default : 9200)--elasticsearch-index=[index_name]
Name of the index to use--elasticsearch-type=[type_name]
Name of the document type to use
- Project's wiki
- Description of events fired by phantomas core
- Description of helper functions available to the browser in window.__phantomas
- Best Practices for Speeding Up Your Web Site (by Yahoo!)
- Web Performance Best Practices (by Google)
- Writing Efficient CSS (by Mozilla)
- Planet Performance - news and views from the web performance blogosphere
- Performance of 3rd Party Content (by Steve Souders)
- Profiling CSS for fun and profit. Optimization notes.
- phantomas – PhantomJS based, modular web performance metrics generator (an article for Performance Calendar)
- How to Measure Frontend Performance With Grunt
- Know Your Engines: How to Make Your JavaScript Fast (by David Mandelin from Mozilla)
- Velocity Conf 2013 Workshop: Avoiding Web Performance Regression (by Marcel Duran from Twitter)
Use grunt to automate daily dev tasks, including your's application web performance, via these great tools: