JBoss AS in production environment

[agolebiowski]

Hi,

I found very interesting conversation about JBoss AS on
http://groups.google.com/group/Omaha-Java-Users-Group/browse_thread/thread/c3069ffeebd1a808

What is not true in this post??
----------------------------------------------------------------------------
Dave,

I have seen people use JBoss in production. It has come a long way.
Here are some strength and weaknesses IMHO:

Strengths:
It's free to deploy as many as you want
It's fairly easy to get it up and running
It usually supports the bleeding edge specifications
JBoss works well with other open source technologies like Tomcat,
Jetty, MySQL, Hibernate, Seam, etc.

Weaknesses:
To cluster the JBoss app server is no easy feat. documenation lacks
substance
No support except for JBoss Consulting and development community
Lack of a good Admin console, this is a big issue for me, I'd like to
see the configuration and runtime information in an easier manner,
viewing MBeans is not my idea of a complete management console
the LGPL license could be an issue when dealing with state and
governmental agencies. There's something in the GNU license that
states against governmental use.
JBoss Support List price is similar to those of IBM, BEA and Oracle
Marc Fleury, the founder of JBoss, is no longer at JBoss with RedHat
purchasing them out
Recently acquired Arjuna for its Tx Manager, Transaction cannot span
JVMs (Tx context only propagated within VM)
Incapable of automated recovery for XA transactions after a server
crash, and how widely is Arjuna TM used for Mission-Critical
applications? :)
JBoss JMS *replaced* in release 4.2, 5, I don't know how it compares
JBoss 5 finally moving to HTTP Primary/Secondary Replication, AFAIK No
special support for DR/HA architectures
Inefficient resource management (sockets, threads, etc.)
Lots of open ports - 1098:JNDI, 1099:JNDI, 4444:RMI, 4445:RMI,
8009:AJP, 8080:Web, 8083:HTTPClassLoader, 8093:JMS
Sockets left open after application processing complete, brittle under
high loads
Serious issues in app deployment showing race conditions where two
nodes could deploy things in different orders based on farm
deployments to different nodes "at the same time"

As you can tell, I'm sort of Anti-JBoss for an enterprise-ready
environment. For a small mom & pops shop, it's sufficient. :)

Btw, disclosure, I work for BEA Systems, and I do run JBoss with
Tomcat and Apache at my home in my basement on a Intel Pentium II
machine. :)

Regards,
Sheng

On Nov 1, 8:42 am, "David Kerber" <[e-mail]> wrote:

-------------------------------------------------------------------------------

[slaboure]

To cluster the JBoss app server is no easy feat. documenation lacks
substance

All AS docs: http://www.jboss.org/jbossas/docs/index.html

Including an ~110 pages clustering guide: http://www.jboss.org/file-access/default/members/jbossas/freezone/docs/Clustering_Guide/beta500/html_single/index.html

What else do you need?

No support except for JBoss Consulting and development community

Wrong, our business IS to provide quality support and maintenance, 24/7, up to 1h response time, follow-the-sun model.

Lack of a good Admin console, this is a big issue for me, I'd like to
see the configuration and runtime information in an easier manner,
viewing MBeans is not my idea of a complete management console

Customers can enjoy an amazing management software, fully agent-server based called JBoss ON, with advanced monitoring features, clustering capabilities, etc.: https://docs.jbosson.redhat.com/confluence/display/JON2/Home

the LGPL license could be an issue when dealing with state and
governmental agencies. There's something in the GNU license that
states against governmental use.

Wrong. Show me where in the text of the LGPL? Gov agencies are great users of our software, including the FBI, NSA and the Defense.

JBoss Support List price is similar to those of IBM, BEA and Oracle

?!? show me the price list, again, this is very wrong.

Marc Fleury, the founder of JBoss, is no longer at JBoss with RedHat
purchasing them out

So? Alfred Chuang is no more with Oracle, what does it mean?

Recently acquired Arjuna for its Tx Manager, Transaction cannot span
JVMs (Tx context only propagated within VM)
Incapable of automated recovery for XA transactions after a server
crash, and how widely is Arjuna TM used for Mission-Critical
applications?

Hum, all of that is technically very wrong. Arjuna TM used to be HP's TM and was the first Java TM to be both JTA and JTS compliant, it is based on about 20 years of TX engineering out of Newcastle, England. If you can prove just ONE of the points above, please do ASAP.

JBoss JMS *replaced* in release 4.2, 5, I don't know how it compares

Yes, we decided it was better to start on a new architecture to support extended clustering features. Now, JBoss Messaging features better clustering features than WebLogic implementation.

JBoss 5 finally moving to HTTP Primary/Secondary Replication, AFAIK No
special support for DR/HA architectures

Look at the clustering documentation. We have prim/sec replication for quite some time. Our ex-bea customers now using jboss largely prefer our implementation and its performance.

Look, I'll stop there, this is complete FUD from BEA, if they could simply prove anything they say, that would be great. The fact that they need to FUD Java user groups speaks for itself.

Onward,


Sacha Labourey

[agolebiowski]

Thank you for very fast replay:).

[agolebiowski]

Thank you for very fast replay:).

sorry -> reply not replay :)

[anil.saldhana]

I want to ask a simple question to Sheng.

Is BEA is so enterprise centric that it is not suitable to be run in his basement or its fee bit him (in spite of being an employee) that he had to go the FOSS route?

I guess BEA has to resort to FUD on mailing lists (JBoss folks do not do it) because JBoss is eating their lunch/bonuses.