By woodjr on May 09, 2007
486. Could this simple number be stealing the show at the world's largest developer conference?
Maybe. It's getting some help, though. 100, 3, 5, and 2.5 million are also important numbers. Specifically, I'm talking about:
|486||Time (in msec) it took for GlassFish v3 to launch in a keynote demo|
|100||Max memory footprint (in KB) for the GlassFish v3 kernel|
|3||As in "GlassFish v3"|
|5||As in "Java EE 5"|
|2.5 Million||Number of GlassFish downloads to date|
But again, I think 486 is the one that really caught people's attention. App Servers just don't start in half a second. And they sure don't fit into 100KB. So what's going on here?
Well, in fairness, only part of GlassFish v3 is starting in that 486ms. But it's the core part. Everything else can (and is) only loaded when it's actually needed. So you only pay (in memory and initialization overhead) for what you use. It's kind of like we put Sparky (our beloved mascot) on a dynamic diet so that he automatically shrinks or grows to exactly the size you need.
That really could be a game-changer, redefining how we look at App Servers and where we use them. In the conference sessions, for example, many have asked about the possibility of running GlassFish v3 on small devices like phones. And in the blogosphere, Adam Bien wonders whether it could become common to embed GlassFish v3 inside rich client applications. They're interesting possibilities (and all indications are that both could work well).
Let's not forget the last couple of numbers in our list. Java EE 5 makes similar reductions for the consumption of developers' time that GlassFish v3 will do for the consumption of machine resources. It makes development of EJBs and other EE components much simpler and faster by using quick and easy annotations to replace what previously required a lot of boilerplate code and configuration. And 2.5 million downloads demonstrates just how much traction existing GlassFish releases have achieved. So while we do have to wait for v3 to reach production readiness, we have strong options in the meantime including GlassFish v1 (available in a production-quality final release for over a year and used in high-volume sites such as Wotif.com) and GlassFish v2 (which adds new features such as clustering and scripting language support and has just reached Beta 2--making it very close to a final production-quality release).
Yes, I'm biased because I work at Sun and specifically on the GlassFish project. But I really do think that GlassFish (and specifically the modular v3 architecture) has really been grabbing some attention here at JavaOne. And if you aren't here to see Jerome's demonstration in person, don't worry--you can get similar information from his recent v3 screencast.