[CMC] CESG Poll Results - SOIS Documents
Neil Dissinger
neild at aiaa.org
Mon May 28 09:54:24 EDT 2007
CESG E-Poll Identifier: CESG-P-2007-05-001: Determination
of Standards Track Selection of SOIS Draft Specifications
Results of CESG poll beginning 24 May 2007 and ending 16
May 2007:
Recommended Standards (Blue): 3 (50%) (Peccia, Taylor,
Gerner)
Recommended Practice (Magenta): 1 (16.67%) (Barkley)
Experimental (Orange): 2 (33.33%) (Shames,
Durst)
CONDITIONS/COMMENTS:
Peter Shames (Experimental (Orange)): These
documents, at their present state of maturity and with
their current content, do not meet the criteria for CCSDS
Recommended Standards. A Reference Architecture document,
when it is produced, would likely meet the criteria for a
Magenta Book. These books provide a loose definition of an
abstract service interface, but do not define it in either
any concrete form nor in a form which can be translated
into an interoperable implementation. In order to get
these documents, which do represent a significant effort,
out into the public view for broader scrutiny, I would
recommend that they be distributed as CCSDS Orange Books.
Nestor Peccia (Recommended Standards (Blue)): This
discussion should be done at Agency level when reviewing
the RBs as future recommended BBs.
It is better to decide the color "a posteriori" (if RIDs
are generated accordingly), because the raised RIDs are
consolidated / discussed / agreed at Agency level. A vote
"a priori" (sometimes) includes the particular flavour /
interest of ADs / DADs, and are irrelevant "vis-a-vis" the
consolidated position of their Agency.
Erik Barkley (Recommended Practice (Magenta)): As
presently constituted, the set of books are close to a
standards track (Blue books), and represent a very good
start towards such. However, in my opinion, they are still
too abstract to promote interoperability (for example, the
datatype of a device identifier is not defined). I do not
think they quite represent an experimental track so much
as they represent a distillation of the types of services
that occur for on board spacecraft systems. As presently
constituted, the books strike me as detailed material most
closely related to best practices. With further fleshing
out toward more on-the-wire protocol definitions, etc., I
will vote for Blue.
Chris Taylor (Recommended Standards (Blue)): The
present SOIS documents proposed for Agency review consist
of a Green book which includes an overall architecture,
derived communication services and a strategy for arriving
at a set of accompanying protocols to be developed, in
many cases, by external expert teams; together with a set
of service definition Red books.
The strategy of releasing service specifications in
advance of protocol specs may be unusual for CCSDS
(although consistent with ISO which publishes service
definitions independent of protocol specifications) and
has been taken as the most effective and probably only
practical way forward. For example, the subnetwork service
definitions will be used in conjunction with different
underlying data links each with its associated, and
dedicated, protocol. The data link protocols will not be
developed concurrently but all require a single service
specification to be available. This is different to
previous CCSDS recommendation where a service is supported
by a single underlying protocol.
Another peculiarity of SOIS is the extremely broad scope
of work that needs to tackled including, for example, Plug
and play, an area that is still not fully investigated.
Having an agreed framework available as a guide for
further development will allow groups from divergent areas
to harmonise their efforts and agree common as common way
forward. This is already proving its validity with the
announcement of the international SpaceWire working group
to develop SOIS compliant protocols and the preparation of
a CCSDS AMS subset as the protocol for the SOIS message
exchange.
It is of course fully acknowledged that the final purpose
of CCSDS Recommendations is to enable interoperability and
without the specifications of protocols to support the
present service definitions, this will not be achieved.
The SOIS strategy to enable interoperability is a two step
process: release of Green and Draft Red books to provide
the solid platform necessary for protocol development and,
once the protocols have been completed, propose them for
adoption by the CCSDS as Blue Books.
As already stated in the SOIS plan of work, it is intended
to hold the service definitions at a Draft Red book stage
until protocol specification have been completed,
including at least one data link protocol. It is not
therefore proposed to request Blue status until associated
protocols are available. It is, however, important that
the draft Red books are released formally into the
community to provide the necessary framework and CCSDS
guidance and Agency level.
The SOIS area believes that the outlined approach is
consistent with CCSDS procedures for a Blue Book track and
furthermore has the necessary Agency support to proceed.
Bob Durst (Experimental (Orange)): The SOIS
documents have two potential goals: interoperability
and/or application portability. Both of these, when the
standards are complete, mature, and internationally agreed
upon, should result in Blue Books. However, they are very
different goals.
Interoperability (in the way that CCSDS views it) is the
stated goal, and is established through protocol
specifications. None of these documents have protocol
specifications, however, and until they do they cannot be
reviewed to determine whether they actually result in
interoperability. At a point where protocol specifications
become available, the degree of international consensus
should be used to determine whether they are blue or
orange.
Portability is another goal that is an important and
useful target that could be addressed by the SOIS working
groups and the service specifications that have been
produced. But here, too, the "meat" is missing. A
portability specification allows application portability
across spacecraft and spacecraft operating systems (for
example, the POSIX interface is a portability
specification based on the Unix system interface, and
allows applications to be ported across many different
underlying hardware and operating system implementations).
The thing that must be reviewed in a portability
specification is the Application Programming Interface and
Language Bindings for one or more programming languages.
The service specifications that have been produced are an
abstract representation of that interface, but until
language bindings are produced, one cannot actually review
the proposed standard to determine whether it will
actually result in application portability.
So something is missing in either case: either a protocol
specification or an API and at least one language binding.
When SOIS decides (interoperability or portability --
although I believe that the default decision is
interoperability), then it should produce the remainder of
the specifications necessary for reviewers to determine
whether the goal has been achieved. I provided what review
comments that I could on the specifications as I received
them. However, I cannot assess whether they will indeed
promote international interoperability, because no
protocol is specified.
As it stands, the SOIS documents are both incomplete and
immature, so I believe that international consensus will
be (or at least SHOULD be) difficult to achieve. Until
each service specification is accompanied by a protocol
specification, I don't believe that can or should be
approved as a blue book, and I don't think that meaningful
agency review is possible.
Total Respondents: 6
All Areas responded to this question.
SECRETARIAT INTERPRETATION OF RESULTS: Results are
superseded by agreement
to release
service specifications as
draft Recommended
Practices.
PROPOSED SECRETARIAT ACTION: Initiate CMC poll for
authorization
to release of
documents
* * * * * * * * * * * * * * * * * * * * * * * *
More information about the CMC
mailing list