Closed Bug 1222205 Opened 9 years ago Closed 9 years ago

discuss: tools for measuring page performance?

Categories

(www.mozilla.org :: Analytics, defect)

Production
defect
Not set
normal

Tracking

(Not tracked)

RESOLVED WORKSFORME

People

(Reporter: jbertsch, Unassigned)

Details

Hi Team-

What do you think are the best tools for bench marking and measuring web page performance?

I'm under the impression that http://www.webpagetest.org/ is becoming the industry standard.  What do all of you think?  What do you think are the most useful measurements provided by http://www.webpagetest.org/ ?

Alternately, GA also provides site speed/performance data.  What do you think of the suggestions to improve speed that GA offers for /new?  https://www.google.com/analytics/web/?authuser=0&pli=1#report/content-site-speed-suggestions/a36116321w64087799p65789850/%3F_u.date00%3D20151025%26_u.date01%3D20151031%26_.goalOption%3D1%26_.useg%3Dbuiltin1/

Are there other tools you prefer?  What do you think are the most useful/helpful ways to define page performance?

I'm working with Ben and Gareth on creating www.mozilla.org dashboards and I want to make sure that performance measurements for top pages are included in the dashboards.

Let's discuss this here and, if you want, in our next Thursday meeting.

Thanks,
Jen
I would vote for not using GA as having a separate tool probably makes the results more honest and real.

It would be nice to find a tool that can:

* web based saas tool
* do speed tests from various points in the world
* scheduled speed tests
* results of tests over time
* alerts when speed tests reach a specific threshold
* simulate slower connections
* simulate mobile and desktop devices

Or some sub-set of the ideas above.
webpagetest is awesome and using it in conjunction with setting a performance budget and constantly checking it with something like https://github.com/tkadlec/grunt-perfbudget/ is a win.

There are also a lot one can do with regards to fonts, as well as the usual image compression etc.

JS and CSS itself can be big wins so, cleanup of these will be a huge help. Tools such as uncss come in real handy here
If we're just talking about tools to measure performance, we also have New Relic at our disposal. Google's Page Speed Insights are also a good measure.

It's great to see us start to take performance more seriously (i.e through monitoring, dashboards etc). What I would really like to see is the idea of performance as a top-level goal when we design and build new pages. Critical pages should be built to a strict performance budget, which needs to be stuck to over time (including the running of and overall progression of growth type tests). If we did this from the outset, then we should have less technical debt to pay off overall :)
(In reply to Alex Gibson [:agibson] from comment #3)
> If we're just talking about tools to measure performance, we also have New
> Relic at our disposal. Google's Page Speed Insights are also a good measure.
> 
> It's great to see us start to take performance more seriously (i.e through
> monitoring, dashboards etc). What I would really like to see is the idea of
> performance as a top-level goal when we design and build new pages. Critical
> pages should be built to a strict performance budget, which needs to be
> stuck to over time (including the running of and overall progression of
> growth type tests). If we did this from the outset, then we should have less
> technical debt to pay off overall :)

Yeah and when new pages are designed and assets are created, we should try to ensure that they are not over some threshold of load time. It's pretty easy to over-design a page with too many assets or complex interactions that add to the speed of the page.
Both the Summary and Details tab on webpagetest.org offer really useful information. Specifically, First Byte and Start Render on the Summary tab (neither of which we scored particularly well on).

Also interesting are the waterfall and content breakdown (bytes) charts (again, on the Summary tab). Did you know 60% of our home page's weight is in fonts, totaling about 600kb? Ouch.

I know this bug is about measurement, but we should look in to both reducing our font stack and using a font loader to speed up visible text.

(In reply to Alex Gibson [:agibson] from comment #3)
> It's great to see us start to take performance more seriously (i.e through
> monitoring, dashboards etc). What I would really like to see is the idea of
> performance as a top-level goal when we design and build new pages. Critical
> pages should be built to a strict performance budget, which needs to be
> stuck to over time (including the running of and overall progression of
> growth type tests). If we did this from the outset, then we should have less
> technical debt to pay off overall :)

+1000
:jpetto Totally agree about the fonts. There is a lot we can gain for optimizing there.
Status: NEW → RESOLVED
Closed: 9 years ago
Resolution: --- → WORKSFORME
You need to log in before you can comment on or make changes to this bug.