FT.com - 11 facts

October 2016

We switched on the new ft.com as the default for our customers today.

Here’s eleven interesting software engineering facts about it.

++

1/ It is served over HTTPS, and all the *.ft.com assets are served over HTTP/2.

2/ There’s no pre-production environments - just localhost, a build pipeline and production. Boom.

3/ The presentation tier comprised of around 80 independent services, all orchestrated to feel like a single, coherent user interface.

4/ It’s shipped a few hundred times a week.

5/ The bulk of the code & tooling is written in node.js.

6/ Infrastructure is a mix of Heroku, AWS (mostly Lambda, Dynamo, ElasticSearch, Kinesis) with a whole bunch of supporting SaaS providers.

7/ There’s a MVT framework built in to the stack, so each feature can be modified for a user or cohort and it’s impact measured against a control group.

8/ One team owns all aspects of the request chain, from the user's browser all the way to the application (DNS, CDN, routing, presentation tier, hosting etc.).

9/ It’s a fast, progressively enhanced website, and we understand the consequences of it becoming slower.

10/ We removed all the routing cruft from the URLs and rely on HTTP Vary and friends to server different flavours of the site to different people.

11/ The UI is built upon a lovely in-house component framework.

---

All built to distribute the FT's great journalism around the world - try it for a few weeks for a $1 or £1 or €1.