Simon Hearne, Technical Consultant
The setting could hardly have been better. Barcelona was a welcome change from the cold and damp that was England in November, even if it wasn’t quite as convenient as Velocity 2013 in London.
If you’ve attended Velocity before, you’ll know that there’s always a packed schedule, and this year was no exception. After the morning plenaries the sessions break off into different streams, leaving you with a few difficult choices.
It also left me slightly nervous. I was speaking (alongside Andy Davies) at 4pm on day one, and I couldn’t help looking at what else was going on at the same time. Presenting to a big crowd would be fine – talking to a few stragglers who’d got lost on the way to the coffee machine would be downright embarrassing.
In the event, I needn’t have worried. Our talk on third-parties was both well attended and well received. It also felt like we had touched a nerve – a lot of people are struggling to deal with third-party content. No matter how much effort you put into developing a great website, letting third parties on to it can undo all of your work on performance.
If you didn’t make it on the day, you can find the slides here: http://velocityconf.com/velocityeu2014/public/schedule/detail/37181
Andy Davies (right) and me (left) during our presentation on the effect of third-party resources on performance
Like our own session, a lot of the other talks were about the web performance challenges people are facing. Questions included:
- How do you deliver responsive images in a high-performance environment?
- What metrics do you use to measure and report performance to different stakeholders?
- How do you scale a development team while managing front-end performance?
- What are new technologies like Service Worker and HTTP/2 going to mean for web performance?
Some trends, such as responsive design, more third-party content and a tendency towards large, image-rich web pages, can get in the way of performance. It was therefore interesting to hear how different people are dealing with them. There were a number of best-practice presentations, discussing what has worked well for organisations that embrace web performance.
- BBC Sport talked about how their “no faff” approach has improved productivity and increased staff retention (as well as allowing them to develop some really cool tools for rapid prototyping).
- The FT discussed an open-source project to compress PNG images to allow low-bandwidth, photo quality images with transparency.
- WL Square demonstrated how to deliver responsive images using the latest technologies (and back-ups for those browsers that are a bit behind the times).
Overall, it was a great experience and one I’d highly recommend. If you can afford to spend an extra day in the general area, it’s also worth attending the WebPerfDays event that normally follows a Velocity conference.
The next Velocity Europe is in Amsterdam on 11–13 November 2015, and I’m certainly planning to attend if possible. Perhaps I’ll see you there…
Paul Bianciardi, Technical Solutions Architect
I’ve been to a few Velocity conferences now, and they rarely disappoint. Over the years, it’s been especially interesting to see how organisations at the leading edge have got more and more sophisticated in how they think about performance. This year, we heard about some innovative ways to deal with the latest web performance challenges.
Here are a few of my highlights:
Budgets were talked about more and more. More standards need to be defined that the site must adhere to in order to ensure you can deliver content as efficiently as possible. For instance, The Guardian have to deliver the core content in the first 14K budget (to make best use of the initial congestion window of ten packets), and this cannot be broken… by anyone.
Responsive web design all too often comes at the expense of performance. It’s all very well having a website that fits the myriad of device viewports in every case, but pointless (from a UX perspective) if it performs badly, e.g. if you are still delivering all the same content, but just hiding it.
The Financial Times described how they have built for “offline first”. This is a different concept from “prefers online”, which is OK if you just have to decide whether you have a connection or not – however, the reality is that a slow connection can deliver a poorer UX than no connection.
Google are pushing secure everywhere (and are actually moving to 100% encryption, including at rest on disk). Whilst this can introduce an overhead that impacts performance, Google (in the shape of Ilya Grigorik) went through configurations that can reduce this. In the case of Google search, HTTPS (albeit because of SPDY support) is actually faster than HTTP (non-SPDY).
Single page apps (a la AngularJS)
This was mentioned a few times, recognising that more and more sites are moving over to this kind of framework, which complicates the current set of performance metrics, increasing the importance of custom metrics and instrumentation.
Best practice from Etsy
Etsy were again showing “how it should be done”, with their bug hunt mechanism and gamification of finding bugs, with leader boards/prizes and fewer dedicated QAs, where everyone is responsible for using early releases and identifying bugs (championing the idea of “we’ve implemented a new feature – go try and break it”).
All in all, this year’s event was a good illustration of how those of us concerned with web performance are always having to adapt to new trends and challenges.