Velocity of web applications

I am just back from Velocity Europe 2011 in Berlin. This is an O´Reilly conference specialist on Web performance and operations. This is the kind of conference populated with a mixture of system administrators and front end developers. Male to female ratio something like 100 to 1 and black t-shirt to business suit ratio 1000 to 1.Emil in talk with Teyo Tyree Puppet Labs.

The last years we in Visma have been moving from a world where we thought mostly of how to make our systems run smoothly in an environment at the customers site.  We still have to think about these things but now we have a whole new set of issues to deal with when also offering SaaS solutions. The performance of a web application is one of the most important non functional requirements and possibly the most important usability factor. To add to the fun it is an explosion of mobile access to Internet applications whether it is from mobile phones browses, apps or tablets. All with reduced processing and network capacities.

We learned a lot including the difference between speed and velocity. I probably have to do my high school physics classes all over since I  know what speed is but forgot that velocity can be defined as Velocity = speed + direction. Other things are how depressingly much that can and will go wrong and the huge difference between just making a web application and making a web application that scales.

The most entertaining parts where the real world experience of performance improvements and incidents. Here we heard some real refreshing stories of people that had had serious problems and where willing to share the horrid details of what happened and why. I especially liked Tim Morrow from Betfair described how it felt to see the user comments on their support site when having 15 seconds load time to place a bet. And how they simply gave a promise to cut that down to three seconds. Well, now they are down on four but I bet they manage three in the end.

A couple of interesting facts. I listen to Joshua Buxby about the performance KPIs in business scenarios. Real word observations from retailer´s with all kind of portals including full web sites, mobile sites and apps shows that even mobile users prefer the full site. A lot of users are sent to the mobile site, but chose themselves to go to the full site even with mobile or IPad. Especially when it comes to actually buying stuff opposed to just browsing. That hurts a bit since I always advocate for mobile optimized m. sites. All this given us some interesting challenges with respect to performance on the device.

Interesting also that for sites that offer “real” functionality and not just information the iPad rules and have far more buy operations than mobiles. Considering the number of smart phone owners versus IPad owners that is pretty amazing. But considering the demographic profile of iPad users it is maybe not that a surprise after all. (it is the top of Christmas wish list for my wife ( not on her wanted list, but on my give list))

Apart from the real world experience talks we come home with knowledge about a new set of tools like the Google Page Speed and the counterpart Apache module mod_pagespeed.