[orep-wg] WS-ReplicaCatalog specification

Sat Aug 6 04:57:14 CDT 2005

hi rob,

very good! this was not what i understood from ann.
and if you do speak of a WS-ReplicaCatalog interface,
this implies standardization at least to me. otherwise
it is simply a WSRF interface on top of RLS, so excuse
my confusion.

i actually did read the document carefully ;-)

to be constructive, what someone (probably this group) 
needs to do - upon which the rest of OGSA data is relying -
is a WS-ReplicaCatalog interface as you correctly mention
in the document. it needs to be done such that
it fits seamlessly into the puzzle of the OGSA design,
and in this doc i can see no mention or sign of that.

i actually don't agree that it is a spec that anyone could
implement - it does not fit well on top of our existing catalogs
for instance. let me delve a bit deeper technically:

there is a schema defined in the document for how the objects
defined are tied together. this also specifies the implementation
to a large extent, which i believe is the wrong approach. 
i am referring to the diagram on page 8.

i believe that 
 - there should be no such schema given at all, 
   just an interface specification

 - we need to address the specific issue 
   of data replication, not just file replication
   and this has to be reflected well in the design

 - it is not a well-designed schema for the 
   purpose at hand since it is too generic 
   for a replica catalog functionality. it 
   is basically a mapping catalog - there are
   'strings' which may have attributes and 
   which may relate one to another. i believe
   we must be more specific for it to be useful.

 - the interface needs methods that address 
   clearly the creation of data entries, of
   replicas thereof, and also a unique ID 
   as it is present in the RNS and GFS specs.
   this notion is completely missing. stating that
   a GUID may be just one of the strings in
   the mapping catalog is missing the point
   of why a unique ID was defined in the first place.

 - some of the attributes need to be predefined
   for a replica catalog to be semantically useful.
   like creation date, lifetime, etc. there are
   system attributes and user-defined attributes,
   and i'm not sure that user-defined ones really
   belong into a replica catalog interface.

so my question to you: do you intend to do a
WS-ReplicaCatalog spec or do you just intend to put
a WSRF-compatible interface on top of the RLS?

if it is the latter, i suggest you drop the mention
of WS-ReplicaCatalog in the document.
if it is the former, i am happy to participate and then
we need to declare this a proper standardization effort.
we will however proabably need to start from the
existing notions in the other GGF teams and then
see how we should really define the interface,
and map it to existing implementations.

best,

peter

________________________________

	From: Robert Schuler [mailto:schuler at isi.edu] 
	Sent: 05 August 2005 19:32
	To: Peter Kunszt; orep-wg at ggf.org
	Cc: annc at isi.edu; Robert Schuler
	Subject: RE: [orep-wg] WS-ReplicaCatalog specification

	Hi Peter,

	Thank you for taking the time to respond, but did you carefully
read the document? I do not think your comments reflect the actual
contents very well - in my opinion. First, this is in no way presented
as a "GGF standardization document" which is made clear in the opening
paragraph. It states that the document is intended to "generate
dialogue" and "for review and discussion" through the GGF. It makes no
mentioned that we are submitting this document for immediate
consideration for standardization. We humbly seek to begin a discussion.

	As for your explanation regarding what a standardization
document should contain, I hope you do not think we do not understand
that already. When I come across a document that does not match my
expectation of a particular format, I have two options: (1) I can assume
that the person does not know what they are doing, or (2) I can realize
that perhaps they are not intending the document to be of a particular
type (e.g., a standardization document). I would like to believe that in
most instances I choose to give others the benefit of the doubt.

	In terms of your more substantive comment, I would like to
suggest that you take a closer look. We have been careful not to define
anything that is necessarily RLS-specific. So, as far as I can tell,
there is nothing to prevent others from implementing this specification.
On the other hand, at merely 25 pages, it is very terse and incomplete
for a real "standardization document", but we already knew that. It
would have seemed presumptuous on our part, if we were to submit a
complete specification. We have "shared" (as I wrote earlier, not
"submitted") a document with the OREP WG for the purpose of discussion,
at a very early stage in the interface's lifecycle. We are not in the
business of forcing our specifications on others, including our user
community.

	I hope this clarifies our intentions for this specification. If
we misled you in some way, please allow us to correct any
misunderstanding. We are still very interested in your thoughts and
comments from a technical perspective and would like to hear more.

	Thank you,

	rob

	________________________________

		From: Peter Kunszt [mailto:Peter.Kunszt at cern.ch] 
	Sent: Friday, August 05, 2005 8:34 AM
	To: Robert Schuler; orep-wg at ggf.org
	Cc: Ann Chervenak
	Subject: RE: [orep-wg] WS-ReplicaCatalog specification

	hi rob and ann,

	without wanting to sound rude or out of place, please allow me

	to tell you that in my opinion the current scope definition of 

	the document and also the introduction is not suitable 

	for a GGF standardization document.

	it cannot be that a standard is tied to a very specific
implementation

	of a very specific product. a standard should be a just the pure
interface

	definition with its semantics defined, so that every group that
intends to

	implement the standard, can do so in a well-defined way. its
scope has

	to be addressing a well-defined set of requirements in a
specific manner,

	and it also has to mention explicitly what it does not define,
and how it

	relates to other efforts.

	currently the document's first few pages read like an
introduction to

	a work plan for the 'WSRF-ification' of the globus RLS, not like
the intro

	to a GGF standards doc.

	just my 2c,

	peter

	ps. other than this immediate general one, i have also technical
comments - coming soon ;-)

		________________________________

				From: owner-orep-wg at ggf.org
[mailto:owner-orep-wg at ggf.org] On Behalf Of Robert Schuler
		Sent: 05 August 2005 02:56
		To: orep-wg at ggf.org
		Cc: annc at isi.edu; Robert Schuler
		Subject: [orep-wg] WS-ReplicaCatalog specification

		Hello OREP WG members,

		Ann and I would like to share with you a draft
specification for a WS-ReplicaCatalog interface. The intention is to
define a WS-RF interface for a Replica Catalog, similar in functionality
to existing replica catalogs (e.g., the RLS Local Replica Catalog
service). As you know, the earlier OREP specification treated replica
information as individual resources and was an extension of the
WS-ServiceGroup specification. While that approach lends itself to
fine-grain manipulation and inspection of replica location information,
it lacks the concept of "bulk" operations that are so important for
scalability. It is also more difficult to digest some of its concepts.
In this new specification, the interface should be conceptually similar
to the interface provided by existing replica catalog services (e.g.,
RLS LRC, and others). We also expect that this interface will be
complementary to the earlier work on the OREP spec, though it does not
depend on it in any way.

		If you find time to review it, we certainly would like
to hear your thoughts, suggestions, and comments.

		Cheers,

		rob