[Sea-sa] Feedback on the current ASL GB Draft

Tue Jun 12 20:11:46 UTC 2018

Hi Peter,

We discussed this in the SOIS teleconference this morning, and the following ideas appeared:

*	What’s good for MOIMS may also be good for PUS, so the abstraction among these very different implementations would likely erase the tendency to adopt either framework, and instead provide a common subset in SOIS that fulfills the objective of interoperability in a flexible way.
*	It may be ambitious to set a tool chain writer to the task of matching SOIS semantic tags with semantically meaningful interface descriptions in MOIMS, PUS, or other framework; however, the tags would at least assist a human in doing the matching.

In any case, this interface at the top of the application support layer has had no experimental tests, and so it cannot be presented as a solution for anything.  One of the reasons for the lack of tests is that on the SOIS side of the ASL interface we are accustomed to moving data without formalizing our view inside the packets, so we have few skills in applying semantic tags for use on the other side of the interface.  The need to pass data through the ASL interface provides a significant use case for semantic tags.

It seems useful at least to mention this line of thought in both the SAWG ASL and the SOIS EDS/DoT green books, because it is an interface in which people will continue to try various solutions.  In the SOIS EDS/DoT green book I would like to include a brief example for each of MOIMS and PUS, so the semantic tags will have a foundation for a future realistic test, and will guide authors of SEDS to choose semantic tags.

Ramon

From: Shames, Peter M (312B) <peter.m.shames at jpl.nasa.gov> 
Sent: Tuesday, June 12, 2018 1:26 PM
To: ramon krosley <r.krosley at andropogon.org>; 'Roger Thompson' <roger.rocketbrain at btinternet.com>
Cc: 'SEA-SA' <sea-sa at mailman.ccsds.org>; 'Jonathan Wilmot' <Jonathan.J.Wilmot at NASA.gov>
Subject: Re: Feedback on the current ASL GB Draft

Hi Ray,

Um, not to put it too delicately, but I think that we all need to be careful about too much "meshing".  Brings to mind being "enmeshed", as in a bunch of whirling gears.

One thing that has become really clear to me in all of our discussions and analysis, including those done by Melvin,  is that while SOIS EDS and MOIMS MAL both have some "overlap" in that they both carry semantic and syntactic information, the domains that they address and the ways that they address them are really very different.  There is probably future point of tangency, fi/when MOIMS services get migrated on-board, but until that time there are really distinct domains.   For very good reasons.  SOIS is fundamentally about resource constrained, real-time, operational environments and SOIS is fundamentally about terrestrial mission operations where resources are not an issue and processing power and bandwidth are abundant.

I think that MOIMS can benefit from any clear SOIS descriptions of spacecraft, components, interfaces, and their interconnections.  That could be input into the MO operations planning and the ops perp part that Roger has described, but I do not think they are ready for that yet.  That is one sort of "meshing", but it really is an infusion of one sort of information into another realm where it can be used.  Made easier, IMHO, by the use of the clear specs provided by the DOT and EDS.

On the other hand, I do not think that SOIS would benefit from being "meshed" with MOIMS, in the sense of being pushed to adopt the SM&C / MAL, etc service framework.  There is no apparent benefit to adopting what is a rather heavy weight, complicated, multi-layer approach to service interfaces.  I think it would bring "mass" without much value for this application, but others may have different opinions.

As for the Yellow Book, I think it is fine as it is.  It does not need "saving", it needs publishing.

For what it is worth.

Peter

From: Ramon Krosley <r.krosley at andropogon.org <mailto:r.krosley at andropogon.org> >
Date: Monday, June 11, 2018 at 5:53 PM
To: Peter Shames <peter.m.shames at jpl.nasa.gov <mailto:peter.m.shames at jpl.nasa.gov> >, Roger Thompson <roger.rocketbrain at btinternet.com <mailto:roger.rocketbrain at btinternet.com> >
Cc: SEA-SA <sea-sa at mailman.ccsds.org <mailto:sea-sa at mailman.ccsds.org> >, "Wilmot, Jonathan J. (GSFC-5820)" <Jonathan.J.Wilmot at NASA.gov <mailto:Jonathan.J.Wilmot at NASA.gov> >
Subject: RE: Feedback on the current ASL GB Draft

Today, I was reminded again of a missing part of the ASL green book content by materials that Jonathan distributed for our next SOIS teleconference.  This reminder comes from Mehran Sarkarati in comments on the details of Richard Melvin’s report on the “overlap” between MO Services and SOIS EDS, and it is summarized by Mario Merri.  Mehran’s comments say that the description of SOIS EDS in the yellow book lacks semantics, only expressing “receive data” and “send data”, while the descriptions (elsewhere) of MO Services “define operations with semantically meaningful arguments”.  Mario’s summary (in my words) says that we don’t need overlap; we need meshing.

These comments are correct, except for the lack of semantics in SOIS.  The situation is an example of how we sometimes “speak past” one another.

We have the opportunity in the ASL green book to show meshing between SOIS and MOIMS, instead of overlap or parallel play.  The capability to mesh interfaces comes from the presence of semantic content in both MO Services and in SOIS EDS.  An exercise is needed here, to compare the MAL terms with the terms that are available in SOIS, to show an example of how the meshing could occur.  If Roger can provide me with one or a few examples of MO Services with semantically meaningful arguments, then I can either find the terms in our DoT, or add them to the DoT, and extend our star tracker example SEDS with semantic tags that enable some form of meshing with the requirements of MO Services.

The result of this exercise could provide material to save the yellow book.

Ramon

From: ramon krosley <r.krosley at andropogon.org <mailto:r.krosley at andropogon.org> > 
Sent: Monday, June 4, 2018 5:30 PM
To: 'Shames, Peter M (312B)' <peter.m.shames at jpl.nasa.gov <mailto:peter.m.shames at jpl.nasa.gov> >; 'Roger Thompson' <roger.rocketbrain at btinternet.com <mailto:roger.rocketbrain at btinternet.com> >
Cc: 'SEA-SA' <sea-sa at mailman.ccsds.org <mailto:sea-sa at mailman.ccsds.org> >
Subject: RE: Feedback on the current ASL GB Draft

I see what you are looking for, but I don’t find it or it’s not in the form that you might expect it in the current conversation in the SOIS working groups.  This is an opportunity for me to try some drafty explanations that might develop into material for the ASL green book:

*       For the general emphasis on EDS:  People seem to find this concept to be a panacea for adapting to a diversity of technologies where there is clearly an interface, but it is an interface that cannot be standardized.  Emphasis on adaptation to a particular technology in the ASL green book can be an informative example, but not an introduction to a normative standard.

*       For subnetworks:  The role of the content of the SOIS Subnet magenta books has been questioned in the last two face-to-face meetings of CCSDS, and an answer has emerged that is due to Richard Melvin’s work on generating device services from SEDS.  First, some background:  During the SUMO conference a few years ago, I asked a Lockheed-Martin engineer what he thought of the magenta books, and he said that they were “incomplete”.  Based on that conversation, I think he meant that they were so general that they could not be implemented in an interoperable way by two different implementers.  Back to the present:  Richard found that the general patterns of the subnetwork magenta book interfaces could be used to map requests from device services to behavior paradigms of devices; for example, he found that the SOIS Memory Access Service (MAS) could model a device on a 1553 subnetwork, and the SOIS Packet Service (PS) could model a non-RMAP device on a SpaceWire subnetwork.  So now there is a SEDS Package File that describes the interfaces in the SOIS Subnet magenta books, and the SEDS for a device on a given subnetwork makes reference to that Package File for its subnetwork interactions.  Each agency has its own implementation of subnetwork interfaces, which are not interoperable with other agencies, and there is little perception of incentive to make them interoperable.  The tool chain generates the software to implement the mapping for a given agency’s subnetwork interface, using the patterns in the SOIS Subnet magenta books.  This is one of the points to come out of the recent work on the SOIS EDS & DoT green book.

*       For support services, alas:  The conversation on support services in the SOIS working groups has been tepid.  For example, the Time Access Service (TAS), it has been suggested, would be better replaced with a device service for a clock or a GPS receiver.  It could be argued that the TAS is at a different level than is the substituted device service, but this discussion has no takers at the present time.  This is one of the points that is the victim of the current focus of discussion in SOIS, which manifested as silverization of the SOIS application support magenta books.  In the future, we may resurrect the application support magenta books in the style of the subnetwork interfaces, but the interest is simply not present at this time (maybe He Xiongwen has this interest)

*       For wireless:  Given the current conversation, it seems to me that a likely path for integration of the wireless working group with the other two SOIS working groups is to test the hypothesis that devices on a wireless subnetwork can be described in a pattern like that of MAS, PS, or something new.  I promised Kevin Gifford to pass this on to him as soon as I could express it sufficiently well.  I think that I’ll have something for Kevin in about a week now, to start that conversation.

So, in summary, I can add material for the missing topics that you listed, in the manner described above, labelled by the qualities of standards that you listed, “got it”, “got a plan for it”, “nothing yet”.  If you see additional avenues that I have missed, let me know.

From: Shames, Peter M (312B) <peter.m.shames at jpl.nasa.gov <mailto:peter.m.shames at jpl.nasa.gov> > 
Sent: Monday, June 4, 2018 3:19 PM
To: ramon krosley <r.krosley at andropogon.org <mailto:r.krosley at andropogon.org> >; 'Roger Thompson' <roger.rocketbrain at btinternet.com <mailto:roger.rocketbrain at btinternet.com> >
Cc: 'SEA-SA' <sea-sa at mailman.ccsds.org <mailto:sea-sa at mailman.ccsds.org> >
Subject: Re: Feedback on the current ASL GB Draft

Hi Ray,

What are you running in, "Skeletoes"?

Actually, what you just stated " The work on the SOIS EDS & DoT green book during the last couple of months has produced some highly detailed descriptions of the process for generating device services from SEDS, which I intend to summarize at a higher level of abstraction in the ASL green book.  I think that will address at least some of your concern." Does not alleviate my concern at all.  On the contrary is it exactly what I had in mind when I said 

5)      Some of the SOIS limitations are due to the "silverizing" activity that occurred, some of it seems due to a shift in focus by the leadership to implementation approaches and EDS.  Regardless, those SOIS functions and features (support, subnet, wireless) that remain in MB and GB materials really should be adequately covered in this document.

IMHO the whole SOIS effort has pivoted around to where the EDS is the ONLY thing that is emphasized.  And yet there are still subnet, support services, and wireless that are on the books, and IN the books.  There is no question in my mind that the EDS and DOT is one of the most powerful, and "real", things that SOIS has defined, but it is not the only thing.  

I really do expect to see these other topics given an adequate description in this document.

Thanks, Peter

From: Ramon Krosley <r.krosley at andropogon.org <mailto:r.krosley at andropogon.org> >
Date: Monday, June 4, 2018 at 1:54 PM
To: Peter Shames <peter.m.shames at jpl.nasa.gov <mailto:peter.m.shames at jpl.nasa.gov> >, Roger Thompson <roger.rocketbrain at btinternet.com <mailto:roger.rocketbrain at btinternet.com> >
Cc: SEA-SA <sea-sa at mailman.ccsds.org <mailto:sea-sa at mailman.ccsds.org> >
Subject: RE: Feedback on the current ASL GB Draft

Hi Peter,

Thanks for this guidance; it’s always helpful to know where the work is needed.

I agree with your observations on the SOIS material; you should see my running shoes (minimal, but not barefoot).  The work on the SOIS EDS & DoT green book during the last couple of months has produced some highly detailed descriptions of the process for generating device services from SEDS, which I intend to summarize at a higher level of abstraction in the ASL green book.  I think that will address at least some of your concern.

Ramon

From: Shames, Peter M (312B) <peter.m.shames at jpl.nasa.gov <mailto:peter.m.shames at jpl.nasa.gov> > 
Sent: Monday, June 4, 2018 2:32 PM
To: Roger Thompson <roger.rocketbrain at btinternet.com <mailto:roger.rocketbrain at btinternet.com> >; Ramon Krosley <r.krosley at andropogon.org <mailto:r.krosley at andropogon.org> >
Cc: SEA-SA <sea-sa at mailman.ccsds.org <mailto:sea-sa at mailman.ccsds.org> >
Subject: Feedback on the current ASL GB Draft
Importance: High

Hi Guys,

Over the last few days I have been reading through the current (or latest) ASL GB draft.   I think that you guys have made great progress and this document is coming together rather nicely.  At this point I am up to page 61, the end of Section 4, or something like 1/3 of the way done.

In the attached version you will see my mark-up notes.  There are a lot of inputs that are simply editorial, and others that are somewhat more substantive.  In the "substantive" column I would put the following:

1)      Most of what is now in Sec 1.1 really, IMHO, belongs in Sec 1.3, Rationale.  This is the expected usage in CCSDS Green Books, but they have been known to vary.

2)      The MOIMS sections that I have reviewed so far. are very good and quite complete.  In fact, they almost seem too complete, but that is another discussion. If everything is done to this level, as it should be to have an "even" appearance, the doc will be huge.

3)      The handling of the interfaces between MOIMS and CSS does not strike me as satisfactory.  The MOIMS MO services must plan for space links, send data over space links, and get data of various kinds from space links.  It just does not seem acceptable to say " telemetry acquisition and telecommand uplink are communications layer functions that do not directly interact at the application layer with MOIMS functions ".  If not MOIMS, then where?  Is there an air gap?  Does some other CCSDS area have responsibility?  In SCCS-ADD parlance it is "applications" that interact with the SLE and CSTS services.  Those application, IMHO, ought to be MOIMS.

4)      The SOIS materials, by contrast, are good, but seem really incomplete.  Notwithstanding that the scope of SOIS is considerably less than that of MOIMS, the treatment of the SOIS functions and services is itself truncated in the extreme.  As much as MOIMS seems a little overblown, SOIS seems totally underinflated.  I think we need to seek parallel construction in each of the sections wherever possible.  See in particular my comments in Sec 2.4.2 and 4.3.

5)      Some of the SOIS limitations are due to the "silverizing" activity that occurred, some of it seems due to a shift in focus by the leadership to implementation approaches and EDS.  Regardless, those SOIS functions and features (support, subnet, wireless) that remain in MB and GB materials really should be adequately covered in this document.

6)      I like the materials in Sec 3.3, but worry a little that we may be putting too much "RASDS 2.0" materials into this document.  There may not really be much choice, because we have agreed to use these extensions and really must explain what we are doing and why, but it is a little concern.

7)      The distinctions in type and representation that are used in Sec 3.3 among unspecified, published, under development, proposed, and "no standard identified" seem, in general, to be not sufficiently distinct to be useful.  I think the important distinction is really only between "do we have a standard" and "do we have any sort of plan for a standard".   Everything else is just so much idle speculation.  Three levels are probably enough: "it exists", "we have a plan for it", and "something could be (might be) here, but isn’t".

I'll keep on plugging away at this.  Do take a look and see if what I have identified seems reasonable or really off base.

Thanks, Peter

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.ccsds.org/pipermail/sea-sa/attachments/20180612/fc55a85e/attachment.html>