How Comet Brings Instant Messaging to Meebo includes insights from Jian Shen, lead software engineer at Meebo, Inc.

Meebo has been very successful in rolling out large-scale Comet-based products, and the article explains their approach, technology issues they’ve faced, as well as thoughts on how the live web changes search engines, metrics, and more.

There are disadvantages, though, to this method. A typical pitfall in developing asynchronous Web applications is failing to recognize that solid-server architecture is crucial to scaling a Comet-style application. Perhaps the biggest surprise most other developers learn about meebo is that its back-end server architecture has been heavily iterated on from the very beginning specifically to address the needs of instant messaging.

For his work, Shen defines a “solid-server architecture” as one that has been tuned to fit the traffic patterns of the application. He elaborates what this is for meebo: “Every application’s traffic pattern is going to be different and your users are going to determine that for you. The big points are: JavaScript and XMLHttpRequest’s for browser compatibility; long-polling rather than streaming or regular polling for real-time messaging with the expectation that our users might have less-than-consistent internet connections; optimized server-side logic for speed, since our traffic is extremely transient, making our real bottlenecks memory and CPU; and lots of stats built into our system, so we can measure and tune to the traffic pattern of our users.”