Web service on VMS

[Jeffrey H. Coffield]

Jeffrey H. Coffield wrote:

The JNI solution is not needed for client Java - it would be needed
for server Java.

Arne

Even if the Java app runs on OpenVMS, which we have a couple, we use the
same connection logic so we don't have two different methods for Java.
One for OpenVMS and one for everything else. A simple benchmark did not
show a significant difference between using TCP/IP on the same node as
using JNI. I suspect that most of the load is generated in converting
data to and from the OpenVMS native format to what Java wants which is
needed in both cases.

We actually have two different connection methods now. One uses a
TCPI/IP service so you get one persistant connection per Java client
(like FTP or Telnet). This allows for the server to send data to the
client in real time and is more compatible with traditional OpenVMS
locking. The other is based on a HTTP post with an XML response. This
requires the whole session management logic but this allows the same
back end logic to be used with either a Java/Swing/WebStart front end or
HTML/Ajax/Jquery.

Jeff

Jeff

 

[Neil Rieck]

[...snip...]

As of April 2009 the number is 45.95%.

Wow, you are correct. Apache is still in the lead but only with 45%

http://news.netcraft.com/archives/web_server_survey.html

Web service and web apps are really two different things.

True. I only mentioned my it because the original post mentioned Java
but didn't say why.

BTW, I am not against Java per se. It's just that a well-written
application written in any high level language will eat up fewer
system resources. I have even done XML solutions in BASIC and C, but
not solutions based upon cool downloadable is that a real word?) XML
like this:

http://h71000.www7.hp.com/openvms/products/ips/soap/soap.html
http://h71000.www7.hp.com/openvms/products/ips/xml/

Neil Rieck
Kitchener/Waterloo/Cambridge,
Ontario, Canada.
http://www3.sympatico.ca/n.rieck/OpenVMS.html

 

[yyyc186]

Hello.
I want to use web-service on VMS.
I know that Java is a good solution (I think it's version 1.5).
Also I am using Basic in most of my code.

I need a full example code of Java using Web-service (on VMS),
that can be used by VMS-Basic code, please.
(Even 3rd-party program is good).

Thanks

You need to get yourself a copy of "The Minimum You Need to Know About
Service Oriented Architecture" for actual coding examples on OpenVMS.
You can get it from your friends at Island Computers

http://www.islandco.com

or from Barnes & Noble

http://books.barnesandnoble.com/search/results.aspx?ATH=Roland Hughes&SRT=R&DREF=1

This book won a 2008 Best Book Award from USA Book News and is
currently a Finalist in the 2009 Eric Hoffer Awards.

 

[Tom Linden]

I don't use Java on the OpenVMS (server) side for many reasons. It
doesn't naturally integrate with existing OpenVMS applications. It can't
connect directly to the OpenVMS database called RMS. It takes a large
amount of memory and is generally a lot slower. We in general have the
Java client connect to an OpenVMS server written in Basic.

We have partially solved that problem. Go to our website.

 

[Tim E. Sneddon]

I don't use Java on the OpenVMS (server) side for many reasons. It
doesn't naturally integrate with existing OpenVMS applications. It can't
connect directly to the OpenVMS database called RMS.

This is something that has been solved with Kednos' Java-to-VMS
Toolkit. With J2VMS it is possible to directly call all OpenVMS
system services and run-time libraries (even has a STARLET
library). I even wrote a short article for the last VTJ about
who to use it.

It takes a large
amount of memory and is generally a lot slower.

That can be true. However, in my experience it's not always the
case if the software is written and configured correctly. Once
a server application is loaded it tends to be quite quick. I've
written messageing software that uses a number of different
mechanisms for transmitting and receiving XML formatted messages.
These have then been unpacked and forwarded to native application
using mailboxes and files. All using J2VMS, RMS and system
services.

We in general have the
Java client connect to an OpenVMS server written in Basic.

To me it's using the right tool for the right job.

I agree completely. It's entirely about what you want to do and
what expertise and supporting software you have available.


Tim.

 

[Tim E. Sneddon]

It is of course possible to write something yourself using JNI, but
that is not so attractive.

J2VMS does something about that. Check it out at:

http://www.kednos.com/Integration/Java

A Java server program should run fine in a few hundred dollars worth
of RAM in an Integrity server. And about as fast as programs written
in any other language.

If it is a 10 or 15 year old small server maxed out with memory,
then latest and greatest Java everything will most likely not
run very well.

I certainly agree with this.

Tim.

 

[Tim E. Sneddon]

J2VMS does something about that. Check it out at:

http://www.kednos.com/Integration/Java

I made a boo-boo with this. Thanks for pointing it out Arne. The
address should be:

http://www.kednos.com/kednos/Integration/Java

Tim.

 

[Bob Koehler]

No, but the statement "Java is most of the time a bad solution" is more
general. I can imagine it might be a bad solution on OpenVMS, although you
didn't support that claim, but to say that it's a bad solution "most of the
time" is ridiculous. Even when there are better solutions, Java is often a
good solution.

It depends on what you're trying to do. I know folks who are
religiously opposed to interpretting byte code. For me Java on VMS
and several other platforms as been a much better solution than I
though it would be. (The JRE on an EV4 is not as slow as a I
feared.)

 

[Lew]

It depends on what you're trying to do. I know folks who are
religiously opposed to interpretting byte code.

It took me a minute to parse that; I thought you meant opposed to interpreting
bytecode themselves, but I see now you meant opposed to running interpreted
bytecode on a JVM.

That would have to be a religious, well, at least a superstitious issue. I
can think of no rational reason supported by evidence to think that running
bytecode is always a bad idea. I don't know VMS (haven't used it in over a
quarter century), but on other platforms Java's performance is comparable to
and even sometimes faster than that of, say, C++ programs.

I don't know about HP's Basic on VMS either, but fifteen years ago I was with
a company where we wrote an interpreter for HP Business Basic that ran faster
on a 25 MHz 386 (with three simultaneous users in 2 MB RAM total and no
floating point hardware) than HP's own compiled version did on the HP 9000
mainframe.

For me Java on VMS and several other platforms as been a much
better solution than I though it would be.
(The JRE on an EV4 is not as slow as a I feared.)

Who wrote that version of Java? Does it do Hotspot?

 

[Arne Vajhøj]

Who wrote that version of Java? Does it do Hotspot?

The JVM for VMS is provided by Digital/Compaq/HP.

It does JIT compilation. They used to call it "Fast VM" when
doing that.

As far as I can tell that JIT compiler does a good job.

An EV4 CPU is from 1992-94.

Arne

 

[Neil Rieck]

The JVM for VMS is provided by Digital/Compaq/HP.

It does JIT compilation. They used to call it "Fast VM" when
doing that.

As far as I can tell that JIT compiler does a good job.

An EV4 CPU is from 1992-94.

Arne

The following isn't a complaint, just an observation:

Last year we were playing with server-side java on our development
platform. We needed to go into a different direction and so never
removed the java stuff. It is amazing to me how many resources the
tomcat server eats up while doing nothing. I'm not sure if the JVM is
constantly doing garbage collection or something else is going on, but
this OpenVMS product does not up to the same quality as other ebiz
stuff like SWS.

Neil Rieck
Kitchener/Waterloo/Cambridge,
Ontario, Canada.
http://www3.sympatico.ca/n.rieck/OpenVMS.html

 

[Neil Rieck]

The following isn't a complaint, just an observation:

Last year we were playing with server-side java on our development
platform. We needed to go into a different direction and so never
removed the java stuff. It is amazing to me how many resources the
tomcat server eats up while doing nothing. I'm not sure if the JVM is
constantly doing garbage collection or something else is going on, but
this OpenVMS product does not up to the same quality as other ebiz
stuff like SWS.

Neil Rieck
Kitchener/Waterloo/Cambridge,
Ontario, Canada.http://www3.sympatico.ca/n.rieck/OpenVMS.html- Hide quoted text -

- Show quoted text -

Sorry for responding to my own post, but I need to add one important
point. Last year we were running Tomcat 3.0 but it looks like Tomcat
3.1 fixes a few problems
http://h71000.www7.hp.com/openvms/products/ips/apache/csws_java.html

NSR

 

[Lew]

Sorry for responding to my own post, but I need to add one important
point. Last year we were running Tomcat 3.0 but it looks like Tomcat
3.1 fixes a few problems

Tomcat 3.1? Looking at
<http://tomcat.apache.org/>
I see that the latest Tomcat is 6.0.18 and the earliest one even listed is
4.1. Going into "archives" I can't find Tomcat 3.1; they list version 3.3.2.

I'm betting that if 3.1 fixes a "few" problems, that 6.0.18 fixes a few more.

 

[Arne Vajhøj]

Tomcat 3.1? Looking at
<http://tomcat.apache.org/>
I see that the latest Tomcat is 6.0.18 and the earliest one even listed
is 4.1. Going into "archives" I can't find Tomcat 3.1; they list
version 3.3.2.

I'm betting that if 3.1 fixes a "few" problems, that 6.0.18 fixes a few
more.

Let me translate:

CSWS_Java 3.1 = Tomcat 5.5.26
CSWS_Java 3.0 = Tomcat 5.5.9

Arne