For Web apps, performance remains an uphill battle

It seems like there ought to be a way to load content as it becomes available instead of having to wait on secondary content which, let's face it, is not what most people come to a site to view anyway. That said, running Firefox with NoScript can do away with some of the lag since it stops JavaScript from executing by default and thus keeps the user from having to wait for subsidiary content to load. On a related note, many software vendors now supply Web interfaces in lieu of native clients for their products, and the Web interfaces are, without exception, inferior: they're slow (even on a gigabit LAN), are less intuitive, and generally don't provide the same degree of functionality, suggesting that the Web browser is not a fully-fledged OS replacement, even after all this time.

As a perfect example, look no further than the InfoWorld site and all the crap it loads when you want to read an article :)

There are significant speed, performance and features limits to web-based AJAX applications. Why not develop platform-native Cloud based applications? These desktop can take full advantage of the underlying platform's full power (such as a Cocoa app for Mac or a C#/.NET app for Windows) and makes RESTful calls to read/save all data - ie. no data is saved on disk. Is cloud computing applications of the future really an all-web world? No, I don't think so. Sometime later, people will start thinking about the raw performance and feature benefits and rethink about desktop apps. Kevin Pan (Open Source Developer) http://ObjectsOnClouds.org - Open Source Cloud Computing Project

It's not a battle. It's a matter of testing and capacity planning. It's not fair to say, "for web apps in general," because that is a generalization. IT teams, developers, and the businesses they serve must (and often do) consider performance as part of the application requirement. To do so means certain performance metrics are targeted for the project, and must be attained and sustained. There are best practices and ample tools for this, including but not limited to load testing, stress testing, regression testing, web performance measurement and monitoring, etc. Depending on the performance levels targeted, applications can be designed to scale accordingly, using code optimization/refactoring of bottlenecks, load balancing of servers, more hardware, etc. From the user perspective, building apps with rich user interfaces can improve real performance, as well as the appearance of performance. For example, a database-driven app that requires data be submitted and then fetches (and refetches) web pages appears slower (and often is, in fact, slower) than a database-driven app that uses AJAX. The AJAX GUI can maintain asynchronous connections to the server, keeping data "live" on the end user's screen, and presenting a desktop-like experience through a browser. Ditto Flash, Silverlight, Java, or native apps pushed through a browser, although I would not recommend those approaches for business apps, as they are essentially fat clients that require downloading of plugins and code. This reduces performance, both real and perceived, and defeats the purpose of what we're talking about here. Give users high-performance GUIs (my vote is for AJAX, as it works with all the leading browsers), have targeted performance metrics, test and tune during development, capacity plan your hardware in the cloud properly, and measure and refactor (if neccessary) after deployment. Richard Rabins Co-Chairman Alpha Software http://blog.alphasoftware.com