Recently I’ve been talking to lots of banks about building HTML5 web trading apps, many of whom are looking to move from a previous generation web-delivered SDP built in Flex (Flash) or Silverlight to an HTML5-based app, or set of apps. It was interesting to see a clear split in their existing platform architectures that seems to almost exactly map to a thick vs thin approach.
The thick approach pretty much treats the web as a container for delivering a traditional application. Many existing .NET trading apps have been ported to Silverlight this way – it usually means sending lots of binary data between the client and the server and relying on decent bandwidth and client processing power to deal with it all. This is often a rapid way to extend an application out to the web, as lots of code can be re-used from existing desktop implementations.
The trouble with the thick client approach is that it doesn’t work well with HTML5 – it’s not really how web apps are built. You are running in the browser’s environment, often with a contended internet connection, and users expect fast loading apps with instant access. Plus HTML5 doesn’t really work with binary data, you really need a good abstract lightweight data model in the browser.
This can cause problems when people try to migrate from Silverlight or Flex to HTML5. If they’ve used the thick approach then it often takes them by surprise that it’s not as simple as just sticking a new web front-end on the existing platform!
I’d be interested to hear how anyone else has managed this migration challenge… I’ll be posting some more thoughts soon on how to design a good platform infrastructure for thin HTML5 clients, and some of the services that are needed to support them.
Filed under: Web trading technology