I have a pet peeve about suggestions that all these exciting Web 2.0 innovations are solely realized within browser-resident Ajax/JavaScript code. For example, consider occasional inquiries regarding a desire to marry Zimbra’s sexy Ajax enduser and administrative UIs with, say, a more mature back-end messaging server. Obviously, one way to respond to such requests is to continue to prove out the Zimbra server’s fault tolerance, scalability, and performance. But I think such requests are more fundamentally misguided.
I haven’t, for example, heard similar calls for decoupling Flickr’s Ajax client from the Yahoo! data center, or for decoupling Microsoft OWA from Exchange. While those Web 2.0 technologies are not open source, they do showcase the fundamental interdependency, just as within Zimbra, between the Ajax client and the backing server.
Ajax applications, of course, tend not to maintain persistent state, and so any update operation typically has to touch the server (via an XML/SOAP request). Similarly, Ajax applications generally have access to more data than they can effectively cache in browser memory (such as your entire mailbox), so queries are also server bound. But more importantly, every Zimbra Ajax innovation has a server component that works in conjunction with the client code. Some innovations are more client centric: Zimlets, ALE, the administration of multiple domains, etc. However, for many of Zimbra’s “UI” innovations, the majority of the smarts are actually implemented on the server-side!: consider search, search builder, saved search, tagging, sharing, VoIP integration, RSS, conversation management, attachment indexing, attachment conversion, the administrative monitoring of Postfix queues, and so on.
Web 2.0 technology is most compelling precisely when the client and server-side code work seamlessly together, typically via fine-grained interactions (over open protocols). Decoupling client from server, then, is unlikely to preserve that experience, unless all of those server smarts are reproduced. In the cases where the Zimbra client has been experimentally used with other servers (which, of course, our open source licensing provides for), it has generally been at the cost of all of the aforementioned innovations!
So while the Web 2.0 excitement will no doubt remain focused on the wizzy new functionality that can now be delivered to the end-user with Ajax, keep in mind that somewhere behind the scenes there are some savvy server programmers that are also making it happen, and if you want new innovative Web 2.0 UI, then you need new innovative Web 2.0 servers too.
Comments are closed.