Installing GlassFish for developing is fairly easy and you are even given multiple options (simple JAR installer, Sun installer, IzPack-based installer, and now zip-based for v3). Installing GlassFish for production certainly needs a bit more care.

Brian Leonard over at the OpenSolaris "Observatory" blog has a recent entry on how to setup GlassFish for production use in a cluster profile. It covers installation from the update center, user setup, domain directory, and SMF service creation.

Also don't forget to read the GlassFish "Installation Guide" and "Deployment Planning Guide" from the standard documentation set.

We just pushed out four new GlassFish Adoption Stories. Three of the stories are from telcos (from the US, the Netherlands and France), the remaining one is a health care company from Canada:

• SFR - Developer APIs, GlassFish-powered - Telco in France.
• T-Mobile, High Availability and GlassFish - Telco in the USA.
• Medavie Blue Cross - Standards Eliminating Vendor Lock-In - Health Care in Canada.
• Pretium Telecom - GlassFish ESB in Telco - Telco in The Netherlands.

When we started GlassFish we used to hear "It's just a Reference Implementation!" so often that I wrote a note on that in June 2006. I think we have made progress since then but I just read Solomon's note on Evaluting JAX-RS Implementations and it has the same misconception, so, here is a reminder...

"A long time ago in a galaxy far, far away", Sun had a group (the "church") doing proof-of-concept References Implementations and Specifications, and another (the "state") doing the commercial products based on those specifications. But those days are long gone and we now have a single team building a single (Open Source) implementation. A Reference Implementation is just an attribute (defined by the JCP), and so is Enteprise Quality (defined by the market), so it is quite possible for an implementation to be both. Like a movie that is in the Top 20 by Revenue and in the Top 20 by AFI... like Star Wars Episode IV: A New Hope.

So, back to Solomon's questions ([1], [2]) - we are building Jersey to be both the Reference Implementation of JAX-RS and the component we will use in GF and several other artifacts. So, go use Jersey, and May the force be with you!

With GlassFish v2 having been out for a couple of months now, a great deal of deployments are happening now. We are highlighting some of these on the Stories blog. This time, it's an entire Java EE 5 Airport Information System (AIS) deployed in production with GlassFish v2. Read about it here.

Among the interesting things in this deployment is the use of almost all the GlassFish features, MySQL and Firebird as the databases, SuSE Linux as the deployment platform, and more. Read about all the details in the detailed questionnaire.

We've previously covered N1 SPS (Service Provisioning System). This new article goes into much greater details about how such a solution and specifically its application server plug-in can help in your larger-scale multi-tier deployment scenarios. The article specifically goes into what a master server and a remote agent are, how a local distributor can help optimize network and firewall communications, how the concept of a component maps to a GlassFish domain or cluster, and how a plan is really a set of instructions requiring a set of variables to be executed. Running a plan can be done as "preflight only" or for real "deployment".

You'll learn how N1 SPS can provision existing installations of GlassFish v2 or carry out entire installations and deployments of SJS Application Server 9.1. This product is really meant for the provisioning of multi-tier deployments. For instance, it lets you install the load-balancing plug-in in the web server tier or manage HADB installation and node management.

In the latest addition to Stories Architect Guillaume Bilodeau shares his experience selecting and later transitioning to GlassFish for the Russian subsidiary of Auchan, one of the biggest retailers in the world (Map, Finances).

In the story and in the questionare Guillaume explains how and why they chose GlassFish over JBoss and others, how they moved existing code to the newer platform from Oracle's application server, what frameworks they use, and finally how the project went from some skepticism over the product choice to happy GlassFish customers.

Stay tuned - we expect more adoption stories in the near future as GlassFish V2's goes final with out-of-the-box enterprise features like custering.

A recent thread at TSS discussed that BEA had certified WLS 10 Technology Preview against the Java EE 5 CTS. That's great news for platform adoption but I'll admit I was unhappy when I read a comment in that same thread that said:

GlassFish doesn't count since the public opinion is to avoid using Sun software except for the JDK...

I find the comment specially grating as BEA is Using GlassFish's Java EE 5 WS Stack in WLS 10. And Sun is using it in Sun's JDK6! And there are more adopters; a non-exhaustive list for JAX-WS includes BEA, Tmax Soft's JEUS 6, Sun's JDK, and more (adoption). The list for JAXB is longer, also including JBoss, Geronimo, Celtix, XFire... (adoption)

I believe GlassFish's Web Services layer is the best in the market; but obviously we need to (continue to?) improve our communication story :-(

Oh well; we will get there eventually - all the annecdotal indicators I see point to strong adoption.

Too many people still believe that a JCP Reference Implementation is always just a "Proof of Concept". Although some are, there are many that are fully production-quality, and, in particular, the Reference Implementation for Java EE has been production quality at least since SJS AS 8.x, so please help us spread the word.

I wrote a blog at Java.Net expanding on the relationship between Reference Implementations and Production Implementations. You may also want to check out our earlier 20 Things You Should Know about Project GlassFish.