Distributed programming in heterogenous platforms

Distributed programming in heterogenous platforms

[Adrian Hoe]

Hello,

I am having a project which is going to be developed using gnat and 
glade. This is a distributed computing involved heterogenous 
platforms. For a start, a small heterogenous platforms will be setup 
with two Sun Solaris workstations and a dual cpu Intel Xeon 
workstation running Linux.

Before bumping my head into the project, I like to gather as much 
information about distributed programming in heterogenous platforms.

Can anyone provide me pointers, experience, websites, sample codes and 
etc. for reference?

Many thanks in advance.
-- 
type Dmitry is new Adrian;           -- Adrian Hoe
                                      -- Remove *nospam* to email

Re: Distributed programming in heterogenous platforms

[Preben Randhol]

On Tue, 24 Sep 2002 01:13:45 +0800, Adrian Hoe wrote:
> Can anyone provide me pointers, experience, websites, sample codes and 
> etc. for reference?

http://libre.act-europe.fr/Software_Matters/main.html

Preben

Re: Distributed programming in heterogenous platforms

[Jim Rogers]

Adrian Hoe wrote:

> Hello,
> 
> I am having a project which is going to be developed using gnat and 
> glade. This is a distributed computing involved heterogenous platforms. 
> For a start, a small heterogenous platforms will be setup with two Sun 
> Solaris workstations and a dual cpu Intel Xeon workstation running Linux.
> 
> Before bumping my head into the project, I like to gather as much 
> information about distributed programming in heterogenous platforms.
> 
> Can anyone provide me pointers, experience, websites, sample codes and 
> etc. for reference?
> 
> Many thanks in advance.

This reference is a bit indirect, but the problem is the same.

I worked on some robotic systems for the U.S. Army. The Army was interested
in developing an architecture allowing modules from various vendors,
implemented using different languages and hardware, to work together.
They developed the Joint Architecture for Unmanned Ground Systems (JAUGS).

This is primarily a message based architecture. The messages are defined
independent of any language, with a specified byte order. Messages can be
separated into two broad classifications; periodic messages and events.

The architecture is defined at http://www.jaugs.org/

Jim Rogers

Re: Distributed programming in heterogenous platforms

[Marin David Condic]

The approach of building heterogeneous systems utilizing message passing in
one form or another is *extremely* popular in military systems. Often, its
just Mil-Std-1553 32 word messages for the sake of simplicity, but I've
personally worked on a number of projects dealing with radio communications
and rocket launching that have been based on message passing in various
forms. (Anybody ever heard of the TADIL-J catalog?) It has its strengths and
weaknesses. Obviously, once you define the message format with high
precision, any computer capable of connecting to the network can play and it
doesn't need any specific OS or programming language. The bad news is that
message catalogs can get quite large and complex and the protocols for
getting units talking to each other with the right timing and information
sometimes gets very convoluted. (O.K. First, I'll tell you to get started,
then you ask this other guy for this data. When you get it, you let me know
and start your power up cycle. In the mean time, I'll be talking to that
other guy over there and asking him to send you blah blah blah... :-)

Message passing systems are so common in military hardware that I'm suprised
that Ada didn't manage to get that quite right. It *almost* works to try to
build message catalogs with tagged records, but since you can't control the
representation sufficiently, you just can't use them to play in that
sandbox. The only way to do it is to resort to kludges that might just as
well have been written in C.

MDC
--
======================================================================
Marin David Condic
I work for: http://www.belcan.com/
My project is: http://www.jast.mil/

Send Replies To: m c o n d i c @ a c m . o r g

    "I'd trade it all for just a little more"
        --  Charles Montgomery Burns, [4F10]
======================================================================

Jim Rogers <jimmaureenrogers@worldnet.att.net> wrote in message
news:3D8F49D8.8000605@worldnet.att.net...
>
> This reference is a bit indirect, but the problem is the same.
>
> I worked on some robotic systems for the U.S. Army. The Army was
interested
> in developing an architecture allowing modules from various vendors,
> implemented using different languages and hardware, to work together.
> They developed the Joint Architecture for Unmanned Ground Systems (JAUGS).
>
> This is primarily a message based architecture. The messages are defined
> independent of any language, with a specified byte order. Messages can be
> separated into two broad classifications; periodic messages and events.
>
> The architecture is defined at http://www.jaugs.org/
>
> Jim Rogers
>

RE: Distributed programming in heterogeneous platforms

[Robert C. Leif]

What would happen if the Ada Distributed Systems Annex were used with
XML messages that were validated by both the sender and receiver with
the same XML schema?
Bob Leif 

-----Original Message-----
From: comp.lang.ada-admin@ada.eu.org
[mailto:comp.lang.ada-admin@ada.eu.org] On Behalf Of Marin David Condic
Sent: Tuesday, September 24, 2002 5:05 AM
To: comp.lang.ada@ada.eu.org
Subject: Re: Distributed programming in heterogenous platforms

The approach of building heterogeneous systems utilizing message passing
in
one form or another is *extremely* popular in military systems. Often,
its
just Mil-Std-1553 32 word messages for the sake of simplicity, but I've
personally worked on a number of projects dealing with radio
communications
and rocket launching that have been based on message passing in various
forms. (Anybody ever heard of the TADIL-J catalog?) It has its strengths
and
weaknesses. Obviously, once you define the message format with high
precision, any computer capable of connecting to the network can play
and it
doesn't need any specific OS or programming language. The bad news is
that
message catalogs can get quite large and complex and the protocols for
getting units talking to each other with the right timing and
information
sometimes gets very convoluted. (O.K. First, I'll tell you to get
started,
then you ask this other guy for this data. When you get it, you let me
know
and start your power up cycle. In the mean time, I'll be talking to that
other guy over there and asking him to send you blah blah blah... :-)

Message passing systems are so common in military hardware that I'm
suprised
that Ada didn't manage to get that quite right. It *almost* works to try
to
build message catalogs with tagged records, but since you can't control
the
representation sufficiently, you just can't use them to play in that
sandbox. The only way to do it is to resort to kludges that might just
as
well have been written in C.

MDC
--
======================================================================
Marin David Condic
I work for: http://www.belcan.com/
My project is: http://www.jast.mil/

Send Replies To: m c o n d i c @ a c m . o r g

    "I'd trade it all for just a little more"
        --  Charles Montgomery Burns, [4F10]
======================================================================

Jim Rogers <jimmaureenrogers@worldnet.att.net> wrote in message
news:3D8F49D8.8000605@worldnet.att.net...
>
> This reference is a bit indirect, but the problem is the same.
>
> I worked on some robotic systems for the U.S. Army. The Army was
interested
> in developing an architecture allowing modules from various vendors,
> implemented using different languages and hardware, to work together.
> They developed the Joint Architecture for Unmanned Ground Systems
(JAUGS).
>
> This is primarily a message based architecture. The messages are
defined
> independent of any language, with a specified byte order. Messages can
be
> separated into two broad classifications; periodic messages and
events.
>
> The architecture is defined at http://www.jaugs.org/
>
> Jim Rogers
>

Re: Distributed programming in heterogeneous platforms

[Jim Rogers]

Robert C. Leif wrote:

> What would happen if the Ada Distributed Systems Annex were used with
> XML messages that were validated by both the sender and receiver with
> the same XML schema?
> Bob Leif 
> 


Clean but painfully slow data communication would be achieved.
XML forces all data to be encoded as strings. There is also a significant
overhead in XML tags being passed with the data. This allows strong
platform independence, but is very inefficient.

For most applications the inefficiencies are not important. The latency
accessing a local or remote database may be orders of magnitude greater
than the XML coding/decoding inefficiencies. For hard real time command
and control systems the overhead of XML can be unbearable. In the robotics
program I mentioned earlier we could not even bear the overhead of full
IP protocol. We needed to send all our data at lower levels in the OSI
stack. We were given a very narrow RF bandwidth and a maximum transmission
rate of 24000 baud, with expected average transmission rates closer to
19.2K baud. In our situation XML would have been unacceptable.

Jim Rogers

Re: Distributed programming in heterogeneous platforms

[Marin David Condic]

Yup. Not to mention all the problems you'll have fitting your average web
page into the 64k-words of memory you get with a Mil-Std-1750a processor.
:-)

Also, with the rapid evolution of Internet standards like XML, a military
program that is going to operate for dozens of years might adopt some spiffy
new standard only to discover that the world races past them to something
else before they've even fielded the system.

MDC
--
======================================================================
Marin David Condic
I work for: http://www.belcan.com/
My project is: http://www.jast.mil/

Send Replies To: m c o n d i c @ a c m . o r g

    "I'd trade it all for just a little more"
        --  Charles Montgomery Burns, [4F10]
======================================================================

Jim Rogers <jimmaureenrogers@worldnet.att.net> wrote in message
news:3D910F02.9090006@worldnet.att.net...
>
>
> Clean but painfully slow data communication would be achieved.
> XML forces all data to be encoded as strings. There is also a significant
> overhead in XML tags being passed with the data. This allows strong
> platform independence, but is very inefficient.
>
> For most applications the inefficiencies are not important. The latency
> accessing a local or remote database may be orders of magnitude greater
> than the XML coding/decoding inefficiencies. For hard real time command
> and control systems the overhead of XML can be unbearable. In the robotics
> program I mentioned earlier we could not even bear the overhead of full
> IP protocol. We needed to send all our data at lower levels in the OSI
> stack. We were given a very narrow RF bandwidth and a maximum transmission
> rate of 24000 baud, with expected average transmission rates closer to
> 19.2K baud. In our situation XML would have been unacceptable.
>

Re: Distributed programming in heterogeneous platforms

[Marin David Condic]

It would probably be too grossly inefficient. Lots of these things are
operating on really tight time budgets and there is usually a need for
highly deterministic behavior. They could have built systems like this using
string data in the Unix/Network flavor but that drastically increases the
amount of data you're sending and increases the time to parse it at the
other end. That's why usually they use raw binary representations. RPCs
might be used in some contexts, but typically that's making assumptions
about what is at the other end of the hose. Binary message passing is just
too efficient, reliable, predictable and flexible to be easily replaced by
something else.

You've got to remember that an awful lot of military systems are built upon
(and will continue to be built upon) really old technology. B-52's first
flew when? 1947? '48? They're still in the inventory and probably will be
for years to come. Environmental requirements (temperature extremes,
rad-hardness, etc.) are often really tough to meet with state of the art
technology. Soldiers don't like it when you bet their lives on new and
unproven technology. As a result, you've got to work with things differently
than is commonly done with Internet-age technology.

MDC
--
======================================================================
Marin David Condic
I work for: http://www.belcan.com/
My project is: http://www.jast.mil/

Send Replies To: m c o n d i c @ a c m . o r g

    "I'd trade it all for just a little more"
        --  Charles Montgomery Burns, [4F10]
======================================================================

Robert C. Leif <rleif@rleif.com> wrote in message
news:mailman.1032912362.1385.comp.lang.ada@ada.eu.org...
> What would happen if the Ada Distributed Systems Annex were used with
> XML messages that were validated by both the sender and receiver with
> the same XML schema?
> Bob Leif
>

Re: Distributed programming in heterogeneous platforms

[Pascal Obry]

"Robert C. Leif" <rleif@rleif.com> writes:

> What would happen if the Ada Distributed Systems Annex were used with
> XML messages that were validated by both the sender and receiver with
> the same XML schema?

This is called SOAP :)

Pascal.

-- 

--|------------------------------------------------------
--| Pascal Obry                           Team-Ada Member
--| 45, rue Gabriel Peri - 78114 Magny Les Hameaux FRANCE
--|------------------------------------------------------
--|         http://perso.wanadoo.fr/pascal.obry
--| "The best way to travel is by means of imagination"
--|
--| gpg --keyserver wwwkeys.pgp.net --recv-key C1082595

Re: Distributed programming in heterogenous platforms

[Pascal Obry]

Adrian Hoe <mailbox@*nospam*adrianhoe.com> writes:

> I am having a project which is going to be developed using gnat and
> glade. This is a distributed computing involved heterogenous platforms. For a
> start, a small heterogenous platforms will be setup with two Sun Solaris
> workstations and a dual cpu Intel Xeon workstation running Linux.

There will be no problem with heterogeneity. GLADE does supports heterogeneous
distribution using XDR protocol. I have built a project with partitions on
Linux and Windows and made some tests on Solaris. This was long time ago...

Pascal.

-- 

--|------------------------------------------------------
--| Pascal Obry                           Team-Ada Member
--| 45, rue Gabriel Peri - 78114 Magny Les Hameaux FRANCE
--|------------------------------------------------------
--|         http://perso.wanadoo.fr/pascal.obry
--| "The best way to travel is by means of imagination"
--|
--| gpg --keyserver wwwkeys.pgp.net --recv-key C1082595

Re: Distributed programming in heterogenous platforms

[Jean-Claude Mahieux]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 992 bytes --]

I would suggest use of CORBA for heterogenous platforms. This is also valid
for a project using different languages.
You can have a look at http://www.topgraphx.com

JC Mahieux.

"Adrian Hoe" <mailbox@*nospam*adrianhoe.com> a �crit dans le message de
news: 3d8f3d90_1@news.tm.net.my...
> Hello,
>
> I am having a project which is going to be developed using gnat and
> glade. This is a distributed computing involved heterogenous
> platforms. For a start, a small heterogenous platforms will be setup
> with two Sun Solaris workstations and a dual cpu Intel Xeon
> workstation running Linux.
>
> Before bumping my head into the project, I like to gather as much
> information about distributed programming in heterogenous platforms.
>
> Can anyone provide me pointers, experience, websites, sample codes and
> etc. for reference?
>
> Many thanks in advance.
> --
> type Dmitry is new Adrian;           -- Adrian Hoe
>                                       -- Remove *nospam* to email
>

Re: Distributed programming in heterogenous platforms

[Victor Giddings]

Adrian Hoe <mailbox@*nospam*adrianhoe.com> wrote in news:3d8f3d90_1
@news.tm.net.my:

> Hello,
> 
> I am having a project which is going to be developed using gnat and 
> glade. This is a distributed computing involved heterogenous 
> platforms. For a start, a small heterogenous platforms will be setup 
> with two Sun Solaris workstations and a dual cpu Intel Xeon 
> workstation running Linux.
> 
> Before bumping my head into the project, I like to gather as much 
> information about distributed programming in heterogenous platforms.
> 
> Can anyone provide me pointers, experience, websites, sample codes and 
> etc. for reference?
> 
> Many thanks in advance.

CORBA is another technology that explicitly deals with platform 
heterogeneity, as well as language heterogeneity. We have a number of 
getting started links on our web site: 
http://www.ois.com/resources/corb-1.asp

-- 
Victor Giddings		mailto:victor.giddings@ois.com
Senior Product Engineer	+1 703 295 6500
Objective Interface Systems	Fax: +1 703 295 6501