> From: Hugh Allen <[EMAIL PROTECTED]> > Reply-To: "Research Triangle Java User's Group mailing > list."<[EMAIL PROTECTED]> > Date: Mon, 21 Jun 2004 16:59:52 -0400 > To: [EMAIL PROTECTED] > Subject: Re: [Juglist] JBoss questions > > Anu, > > 1) I am not using web services, so I can't help with this one. >
Using the JBoss.NET XDoclet tag library you can make any stateless session bean automatically into a web service. JBoss.NET is an integration module with Apache's Axis. > 2) The JBoss business model is that you have to PAY for support and for > training. In my case the cost is prohibitive for both, so I am on my own. This > is the main reason I wish the JBoss documentation was a LOT better. > Check out the wiki (http://jboss.org/wiki/Wiki.jsp) and the getting started guide (http://ads.jboss.org/adclick.php?bannerid=185&zoneid=&source=&dest=http%3A% 2F%2Fwww.jboss.org%2Fmodules%2Fhtml%2Fdocs%2Fjbossj2ee.pdf). I do hope you'll consider coming for the advanced training course (http://jboss.org/services/training/schedule#raleigh-adv). There are also the free forums. (Granted patience is considerably lower for people who ask things on the front page of the website than if you do so on your dime ;-) ) Explain at what cost training wouldn't be prohibitive? I'm rather unapologetic that we make money for what we do. I don't mean to spam...he asked... For case studies about our training/support: http://www.jboss.org/pdf/La_Quinta_Case_Study_FINAL.pdf http://www.jboss.org/company/customers > 3) I am using CMP 2.0 entity beans for ALL of my database interaction. CMP 2.0 > is a beautiful thing, since I just have to write some very simple SQL in the > deployment descriptors. But the tradeoff is less control. My understanding is > that JDBC or JDO could provide more control, at the expense of code > complexity. I am very pleased with CMP 2.0. > Furthermore other setups can give you better speed with CMP. Caching is very important. > You can control JBoss interaction with the database to some extent, depending > on your technical requirements, by specifying loading parameters in your > deployment descriptors. As I understand it, by default JBoss retrieves the key > values for each query and the goes back and retrieves the actual rows of data > later, one by one, as required. However, you can tune this behavior. JBoss > supports: > - load groups (which columns are read) > - read-ahead - on-find and on-load (which rows are read in a single request) > - various commit options that determine whether how data can be cached > (depending on whether JBoss has control over ALL database updates) > - transaction options > - locking options > Add cache invalidation. Other JBoss nodes can tell JBoss when to invalidate the 2nd level cache. Do note that increasingly, Hibernate is becoming our preferred persistence system. In the future Hibernate will be behind our EJB-CMP interface. In the more distant future Hibernate will support EJB3 as an interface directly. Hibernate + JBossCache gives you finer grained control over database/locking and caching. > 4) If you are referring to dynamic loading of EAR files (including WAR and JAR > files), this seems to be one of the BEST features of JBoss. Just drop a new > EAR file in the JBoss server's deploy directory and it will be deployed > automatically. This works great. I haven't used the other major EJB > containers, but I understand they don't have this feature. > The JMX-Console does suck. We're working on something better. There is for now the "http://localhost:8080/web-console"; web-console which is somewhat better but still not up to what I'd like. Its coming though. Generally this is less of a complaint in unix shops where they prefer editing files and more of a complaint in windows shops where they use guis. Configuring load balancing tools is of course a function of how hard it is to configure the load balancing tools (for port 80) such as your hardware or software load balance tool. Really its no harder than straight tomcat there. For RMI load balancing, its a matter of putting "<clustered/>" in your jboss.xml and running "bin/run.sh -c all" the all configuration. Obviously for more than "come up and cluster" there are other options but that will give you load balanced RMI. (client side you'll need to point jndi.properties at port 1100 for HAJNDI instead of straight JNDI) The efficiency of DB connection pooling. Generally we're at a stage that connection pooling is a given. Don't leave home without it. If you mean our particular connection pooling then it has a number of options which trade performance vs safety, but generally has a small performance footprint. Required CPU and RAM. Jboss runs in the default 64 mb heap that the VM comes up with. In fact, I've trimmed 3.2.x to run in under 10mb in the past. JBoss 4.0 requires 128mb heap out of the box but can be scaled back. I've never tested minimum CPU but I've got it running (JbossMail alpha release) on my 733mhz pentium III. I'm not overjoyed with the performance but then again, I'm not overjoyed with the performance of anything else running on it either. I think memory and kernel services like threading are my principal bottlenecks, but I'm on an antiquated linux as well. > Hugh > > At 04:09 PM 6/21/2004, you wrote: > >> Thanks a lot to Phillip Rhodes and Hugh Allen for their invaluable insights. >> Please do comment on the following questions regarding JBoss 3.2.x >> >> 1. Easy implementation of Web Services >> 2. Quality of technical response/support >> 3. Efficient CMP communication with selected database (i.e. is the SQL >> optimized prior to submission or is there lots of interaction with the >> database ) >> 4. Dynamic loading of new files >> >> Thanks in advance >> >> Anu >> >> >> Hugh Allen <[EMAIL PROTECTED]> wrote: >> >>> I have been using JBoss for less than a year, so I am no expert. The >>> learning curve is pretty steep for both J2EE and JBoss. >>> >>> Phillip Rhodes replied separately to your email, and I generally agree with >>> his comments. >>> >>> 1) I believe JBoss supports all of the important parts of J2EE, and each >>> release has more functionality. It has everything I need, and the price is >>> right. >>> >>> 2) I haven't used load balancing or clusters. >>> >>> 3) The JBoss forums have a lot of information. By contrast, the >>> documentation, which costs $10, is not particularly good. It explains a lot >>> about how JBoss works internally, particularly the JMX aspects, but it is >>> not very readable and does not contain a lot of "how to" information that is >>> useful for new developers. >>> >>> Since it is open source, it always seems like the JBoss folks actually >>> expect you to read the source code to answer your questions. After all, >>> that's what THEY do. I have had to do this on at least one occasion, and >>> it's not a great way of figuring out what is going on. >>> >>> 4) The JMX console is not very intuitive. The best source of information to >>> figure out what is going on in the server is the log files. >>> >>> 5) I can't provide any real details on database connection pooling. I am >>> using MySQL with InnoDB tables and performance seems pretty good, but I >>> don't yet have much load on my current applications. >>> >>> 6) I have two Linux web servers running JBoss - both located in the same >>> data center in New York state. >>> >>> The test web server is running Red Hat 9 with User Mode Linux, where the >>> machined is partitioned to support around 20 customers. (See >>> www.rimuhosting.com) Originally, my partition had 128MB of RAM to support >>> Linux, JBoss, Jetty, Apache, and MySQL. I was getting constant out of memory >>> errors. Increasing this to 192MB RAM solved the problem, but this machine is >>> pretty slow, as you might expect. >>> >>> Just set up a new dedicated server running White Box Enterprise Linux 3 with >>> 1GB RAM. This machine is not currently running Apache or Jetty - it uses the >>> embedded JBoss/TomCat instead. This configuration is much faster and >>> response times are quite good. >>> >>> Regards, Hugh >>> >>> At 01:16 PM 6/21/2004, [EMAIL PROTECTED] wrote: >>> >>>> Hi >>>> >>>> I have a couple of questions related to JBOSS 3.2.x. >>>> >>>> 1. Does JBOSS 3.2.x support J2EE 1.3 fully? >>>> 2. Is it easy to configure additional load balancing servers? >>>> 3. Good Usenet group support? >>>> 4. Does it provides application monitoring tool? >>>> 5. How efficient is database connection pooling? >>>> 6. Required CPU and RAM? >>>> >>>> Thanks in advance >>>> >>>> Anu >>>> >>>> __________________________________________________________________ >>>> Introducing the New Netscape Internet Service. >>>> Only $9.95 a month -- Sign up today at http://isp.netscape.com/register >>>> >>>> Netscape. Just the Net You Need. >>>> >>>> New! Netscape Toolbar for Internet Explorer >>>> Search from anywhere on the Web and block those annoying pop-ups. >>>> Download now at http://channels.netscape.com/ns/search/install.jsp >>>> >>>> _______________________________________________ >>>> Juglist mailing list >>>> [EMAIL PROTECTED] >>>> http://trijug.org/mailman/listinfo/juglist_trijug.org >>>> >>>> > > > > --- > Outgoing mail is certified Virus Free. > Checked by AVG anti-virus system (http://www.grisoft.com). > Version: 6.0.708 / Virus Database: 464 - Release Date: 6/18/2004 > > _______________________________________________ > Juglist mailing list > [EMAIL PROTECTED] > http://trijug.org/mailman/listinfo/juglist_trijug.org > _______________________________________________ Juglist mailing list [EMAIL PROTECTED] http://trijug.org/mailman/listinfo/juglist_trijug.org
