Currently set up to get data for 17 "important" URLs as defined by Jess (http://bi.consumeraffairs.com/admin/urldata/url/). These indicate the URLs we have to really keep an eye on.
- still figuring out WebPageTest
- example problem, http://www.webpagetest.org/forums/showthread.php?tid=12012
- one-man project, Patrick Meenan, maintains the server PHP code, the pre-2000 style forum above, and the O'Reilly book ... and has a full-time job elsewhere
- currently, for 17 tests (URLs) it takes around 15 minutes to get results back
- we intend to do an experiment, in an entirely separate AWS zone, entirely separate servers, and run ALL URLs against this server
- we'll let you know
- hopefully
Some features we have in mind:
- More cards in Domo
- Domo Alerts
We currently collect all this data: https://github.com/ConsumerAffairs/bi/blob/master/webpagetest/models.py
For the charts at https://consumeraffairs.domo.com/page/2064862327 (ConsumerAffairs.com/WebPageTest) we practically use a really small part of the data we currently collect; only 3 for those charts, i.e. FullyLoaded metric, response counts per URL and DomElements.
When it comes to future data extraction/collection efforts, this is determined by the previous efforts in extracting the data.
Moar Data. The more valid data, the better. That will probably drive the direction of future Domo cards development. The most important part is that we start actually collecting data.
Have we started? YES! Thanks to this hackathon.