|
Home > Archive > Distributed Computing > May 2007 > RM-ODP question about where to specify an algorithm?
You are viewing an archived Text-only version of the thread.
To view this thread in it's original format and/or if you want to reply to
this thread please [click here]
| Author |
RM-ODP question about where to specify an algorithm?
|
|
| jchludzinski@gmail.com 2007-05-16, 7:13 am |
| In which viewpoint - Computational and/or Engineering and/or
Technology - would I specify the use of a particular algorithm? Or a
particular comms protocol: TCP v. UDP?
---John
| |
| Edward Feustel 2007-05-17, 7:13 am |
|
<jchludzinski@gmail.com> wrote in message
news:1179307903.839648.317280@y80g2000hsf.googlegroups.com...
> In which viewpoint - Computational and/or Engineering and/or
> Technology - would I specify the use of a particular algorithm? Or a
> particular comms protocol: TCP v. UDP?
>
> ---John
>
John,
You would specify technologies to be used in the Technology Viewpoint.
Communications is assumed to be covered in other reference models such as
the OSI model.
The computational viewpoint would be the primary viewpoint used to
specify the algorithm.
There might be some optimization issues that would
be dealt with in the Engineering viewpoint.
Probably some aspects of the "data" might be specified in the Information
viewpoint.
The Enterprise viewpoint might be used to say something about the purpose of
the
algorithm and how it is to be used in the organizational structure.
Ed Feustel
| |
| jchludzinski@gmail.com 2007-05-17, 7:14 pm |
| What if I wanted to specifiy broadcasting messages between nodes and
not a particular mechanism for doing so (e.g., TCP or UDP)? Although,
I'm not sure if the RM-ODP comms mechanisms "operations" (either
"interrogations" or "announcements") or "flows" or "signals" can be
used to describe a broadcast. Where would this go? I assume in the
computational viewpoint?
---John
On May 17, 6:51 am, "Edward Feustel" <efeus...@hughes.net> wrote:
> <jchludzin...@gmail.com> wrote in message
>
> news:1179307903.839648.317280@y80g2000hsf.googlegroups.com...> In which viewpoint - Computational and/or Engineering and/or
>
>
> John,
> You would specify technologies to be used in the Technology Viewpoint.
>
> Communications is assumed to be covered in other reference models such as
> the OSI model.
>
> The computational viewpoint would be the primary viewpoint used to
> specify the algorithm.
>
> There might be some optimization issues that would
> be dealt with in the Engineering viewpoint.
>
> Probably some aspects of the "data" might be specified in the Information
> viewpoint.
>
> The Enterprise viewpoint might be used to say something about the purpose of
> the
> algorithm and how it is to be used in the organizational structure.
>
> Ed Feustel
| |
| Edward Feustel 2007-05-18, 7:13 am |
|
<jchludzinski@gmail.com> wrote in message
news:1179435438.351830.232560@e65g2000hsc.googlegroups.com...
> What if I wanted to specifiy broadcasting messages between nodes and
> not a particular mechanism for doing so (e.g., TCP or UDP)? Although,
> I'm not sure if the RM-ODP comms mechanisms "operations" (either
> "interrogations" or "announcements") or "flows" or "signals" can be
> used to describe a broadcast. Where would this go? I assume in the
> computational viewpoint?
>
> ---John
>
>
> On May 17, 6:51 am, "Edward Feustel" <efeus...@hughes.net> wrote:
>
>
John,
I think it would depend on how you wanted to do it.
I think that a computational view of a Publish/Subscribe Object is probably
the appropriate way to do it. A service registry object is then required. An
object "connection protocol" is required.
The technology view could then include something like the OMG CORBA
Publish/Subscribe.
The engineering view would need to deal with serial communication vs.
multicast, there
might be a multicast object.
You might want to read: "Open Distributed Processing and Multimedia", Blair
and Stefani,
ISBN0-201-17794-3, Addison-Wesley, 1997; I suspect it deals with many of the
issues
in which you are interested in greater detail. See for example Chapter 4 and
8.
Regards,
Ed Feustel
| |
| john.chludzinski@gmail.com 2007-05-21, 1:14 pm |
| Can't use publish/subscribe here; CORBA is way too fat for this
application. This is realtime in nature using a MANET and, to make it
more interesting, we can assume no infrastructure - no wires, no
towers. CORBA/TCP is out of its league here, but what will work is a
form of "controlled flooding" and that means simple broadcasts to
whomever is there. Simple is best! Can RM-ODP be used to describe
such a mechansim (i.e., comms protocol)?
---John
On May 18, 6:43 am, "Edward Feustel" <efeus...@hughes.net> wrote:
> <jchludzin...@gmail.com> wrote in message
>
> news:1179435438.351830.232560@e65g2000hsc.googlegroups.com...
>
>
>
>
>
>
>
>
>
>
>
>
>
> John,
> I think it would depend on how you wanted to do it.
>
> I think that a computational view of a Publish/Subscribe Object is probably
> the appropriate way to do it. A service registry object is then required. An
> object "connection protocol" is required.
>
> The technology view could then include something like the OMG CORBA
> Publish/Subscribe.
>
> The engineering view would need to deal with serial communication vs.
> multicast, there
> might be a multicast object.
>
> You might want to read: "Open Distributed Processing and Multimedia", Blair
> and Stefani,
> ISBN0-201-17794-3, Addison-Wesley, 1997; I suspect it deals with many of the
> issues
> in which you are interested in greater detail. See for example Chapter 4 and
> 8.
> Regards,
> Ed Feustel
| |
| Edward Feustel 2007-05-22, 7:13 am |
|
<john.chludzinski@gmail.com> wrote in message
news:1179760759.903178.79390@x35g2000prf.googlegroups.com...
> Can't use publish/subscribe here; CORBA is way too fat for this
> application. This is realtime in nature using a MANET and, to make it
> more interesting, we can assume no infrastructure - no wires, no
> towers. CORBA/TCP is out of its league here, but what will work is a
> form of "controlled flooding" and that means simple broadcasts to
> whomever is there. Simple is best! Can RM-ODP be used to describe
> such a mechansim (i.e., comms protocol)?
>
> ---John
>
>
> On May 18, 6:43 am, "Edward Feustel" <efeus...@hughes.net> wrote:
>
>
John,
RM-ODP is not a technology that you can use to implement applications. It is
a method for comparing features of different
enterprise/applications/systems/implementations. At least two different
groups use it for describing their implementations: the OMG and The
Telecommunications Information Networking Architecture Consortium (TINA-C).
The Advanced System Networking Architecture (ANSA) project developed some
software to do some of its features in the 1990s under a consortium.
RM-ODP treats communication and its control at a very abstract level since
the OSI Reference Model can be used to compare features of different
communications schemes.
The book by Blair and Stefani may give you more of what you are looking for.
It also occurs to me that you may want to look at implementations of
Distributed Interactive Simulation (DIS). Developed for DARPA in the 1990's,
the methodology and implementations
may be applicable in your case. It used unreliable multicast and was
extremely light weight.
See www.answers.com/topic/distributed-i...tive-simulation for example.
Regards,
Ed Feustel
| |
| john.chludzinski@gmail.com 2007-05-22, 1:14 pm |
| First let me thank you for your time and effort in responding so
thoughtfully.
As for using RM-ODP inappropriately, I don't believe we are. I'm
merely trying to use it to describe a system we've already designed
and simulated. And looking at the mechanisms defined in both the
Computational & Engineering viewpoints, it isn't readily apparent how
to describe a system that relies on simple broadcasts (in our case
802.11 link layer broadcasts - we make NO use of the IP layer). And
no, multicasts groups would be for too onerous to maintain and were
consequently ruled out long ago.
---John
On May 22, 7:30 am, "Edward Feustel" <efeus...@hughes.net> wrote:
> <john.chludzin...@gmail.com> wrote in message
>
> news:1179760759.903178.79390@x35g2000prf.googlegroups.com...
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> John,
> RM-ODP is not a technology that you can use to implement applications. It is
> a method for comparing features of different
> enterprise/applications/systems/implementations. At least two different
> groups use it for describing their implementations: the OMG and The
> Telecommunications Information Networking Architecture Consortium (TINA-C).
> The Advanced System Networking Architecture (ANSA) project developed some
> software to do some of its features in the 1990s under a consortium.
>
> RM-ODP treats communication and its control at a very abstract level since
> the OSI Reference Model can be used to compare features of different
> communications schemes.
>
> The book by Blair and Stefani may give you more of what you are looking for.
>
> It also occurs to me that you may want to look at implementations of
> Distributed Interactive Simulation (DIS). Developed for DARPA in the 1990's,
> the methodology and implementations
> may be applicable in your case. It used unreliable multicast and was
> extremely light weight.
> Seewww.answers.com/topic/distributed-interactive-simulation for example.
> Regards,
> Ed Feustel
| |
| john.chludzinski@gmail.com 2007-05-22, 7:14 pm |
| Ed,
First let me thank you for your time and effort in responding so
thoughtfully.
As for using RM-ODP inappropriately, I don't believe we are. I'm
merely trying to use it to describe a system we've already designed
and simulated. And looking at the mechanisms defined in both the
Computational & Engineering viewpoints, it isn't readily apparent how
to describe a system that relies on simple broadcasts (in our case
802.11 link layer broadcasts - we make NO use of the IP layer). And
no, multicasts groups would be far too onerous to maintain and were
consequently ruled out long ago.
---John
On May 22, 7:30 am, "Edward Feustel" <efeus...@hughes.net> wrote:
> <john.chludzin...@gmail.com> wrote in message
>
> news:1179760759.903178.79390@x35g2000prf.googlegroups.com...
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> John,
> RM-ODP is not a technology that you can use to implement applications. It is
> a method for comparing features of different
> enterprise/applications/systems/implementations. At least two different
> groups use it for describing their implementations: the OMG and The
> Telecommunications Information Networking Architecture Consortium (TINA-C).
> The Advanced System Networking Architecture (ANSA) project developed some
> software to do some of its features in the 1990s under a consortium.
>
> RM-ODP treats communication and its control at a very abstract level since
> the OSI Reference Model can be used to compare features of different
> communications schemes.
>
> The book by Blair and Stefani may give you more of what you are looking for.
>
> It also occurs to me that you may want to look at implementations of
> Distributed Interactive Simulation (DIS). Developed for DARPA in the 1990's,
> the methodology and implementations
> may be applicable in your case. It used unreliable multicast and was
> extremely light weight.
> Seewww.answers.com/topic/distributed-interactive-simulation for example.
> Regards,
> Ed Feustel
|
|
|
|
|