Velocity 2010, Santa Clara, CA – Maximilio Firtman‘s presentation on Mobile Web High Performance presented many useful insights into the intricacies of mobile Web performance and why it matters. Often times, mobile Web browsers are slow, and as mobile hardware platforms are extremely varied, as well as their browsers. The presentation started off with a state of the mobile Web perspective, and some of the caveats with mobile browsers and testing on mobile devices. The second half  of his presentation was filled with useful practices for programming performance into mobile Web applications.

First Firtman comes out and dispels some common myths about mobile browsers. One of these myths is that people are not using their mobile browsers. I would agree with Firtman in that people are using their mobile browsers, but the experience is like that of a socially awkward conversation, you cannot wait to get out of it.  In such cases performance becomes critical, and the application of guidelines such as Firtman’s become a requirement.

Google is one of the main companies which has gone to great lengths to make the Web faster, and more usable. Nobody likes waiting for a Web page, and chances are, if you are browsing the Web on your mobile then you are doing it to serve some purpose that is relevant at that moment, such as finding a restaurant, a map link to a friend’s address, or driving directions. The days of using the mobile Web for fun and enjoyment are not that far off, but we aren’t there just yet.

Firtman drives home one concept that really hit home early in his presentation; the sample space of mobile users is massive, and may even exceed that of laptops and computers. The potential is definitely there for mobile devices to exceed laptop and desktops as Internet connected devices.

One of the major problems that Firtman outlined in his presentation is that there is no documentation for mobile browsers and they are not updateable. This appears reminiscent of the Internet explorer days, where you had a closed, un-extensible browser that was lax in developer features. I recently paged through a pre-release version of Firtman’s upcoming book “Programming the Mobile Web” and it includes some good information about WebKit, Netfront, and other common mobile browsers and frameworks.

One of the great take-aways from Firtman’s presentation was the use of pictograms. Pictograms are pictures that are able to be displayed using the native keymap, and Firtman mentions the emoji set for iPhone in his presentation. It is an effective way to add graphics to a page without the cost of another image render.

Another critical performance metric that I was not aware of until attending this talk is that most mobile browsers can only perform 4-5 parallel downloads, unlike modern browsers which can perform up around 30-60 on average.  This limits the amount of content one should put in a mobile site to make it perform well, and performance becomes even more critical on high-latency wireless networks which many consumers are operating their mobile phones on.

With the increasing sample space of mobile browsers on the Web, developers will need to take mobile performance into consideration more and more as they developer their Web applications and services. Firtman presented many design, performance, and testing considerations that will prove to be useful when developing mobile Web applications.

My take is that HTML5 will be heavily used in mobile applications, and that mobile handset manufacturers will need to provide support for the new standard with increasing importance. HTML5 offers benefits that allow you to style your Web page without adding the overhead of additional assets, such as the additional SVG capabilities.

#velocity2010