RE: [ebxml-dev] Distributive Directories ?

[Zachary Alexander <zack2@cris.com>]

David,

IMHO, there are real problems from a business stand point of defining a
distributed directory the way that you define it.  1) Companies spend a lot
of money acquiring customers.  To place those customers in a common registry
is a non starter.  We see real problems with services like Passport that
allow all participating merchants to share all customers.  I think that
companies will follow the customers.  They will join the registries that
service a specific market segment to address that market segment. 2)  I have
always thought that a common registry really needed a qualifying function.
This function would judge the worthiness of the requestor. Maybe this would
be apart of the negotiation process. One of the potential abuses that could
result is competitors could query the registry for pricing information then
programmatically reduce there pricing.  Now this doesn't really bother me
that much if all participants are aware of the activity. Perhaps special
purpose registries could be set up to satisfy these market needs. In theory,
this would drive market efficiency.  Vendors could reposition and price
products and services based on the spot price of the products and services
and inventory and/or service capacity.

I think that this leads to world that looks more like Farrukh's.  Service
providers and larger organizations will need distributed registries like
Farrukh has described to handle transaction volumes and business continuity.
We are seeing even in the UDDI world a need for private registries or
combination private/public registries.

zack

-----Original Message-----
From: David Lyon [mailto:david.lyon@globaltradedesk.com]
Sent: Friday, April 19, 2002 1:10 AM
To: andrzej@chaeron.com; ebxml-dev@lists.ebxml.org
Subject: [ebxml-dev] Distributive Directories ?


Andrzej,

You've raised some really good points.

Now I don't profess to be an expert in UDDI or WSDL but I've always liked
the general philosophy behind ebXML. Generally speaking the idea of a
registry is a good one, it's virtually an extension of the Trading Partner
information that is kept in most of the old EDI software programs.

Has anybody ever considered a Distributive Directory ?

This is now possible and viable with broadband and offers advantages for
small business over a centralised registry.

The philosophy behind a Distributive Directory is that a company joins an
Exchange and when they do their details are broadcast (name, address,
net-address) to everybody on the exchange.

The details of the new company are stored in the database of all the
companies that are connected.

The result is that within a city or region, everybody can have the contact
details of everybody else and their pricelist/catalog.

One could connect an entire city so that everybody could share everybody
elses information.

With broadband transmission speeds, a 2gig/hertz processor and a 70 gig hard
drive, this seems to be readily achievable. That btw, is the hardware that
the local Plumber can afford.

Surely this sort of technology is on the verge of becoming a reality. There
are some among us who have seen it in operation.

Comments ?

----- Original Message -----
From: "Andrzej Jan Taramina" <andrzej@chaeron.com>
To: <ebxml-dev@lists.ebxml.org>
Sent: Friday, April 19, 2002 2:10 PM
Subject: Re: [ebxml-dev] RE: [EDI-L] Article on ebXML Core Components


> > > Will the registry/repository concept for ebXML eventually merge with
its
> > > counterpart defined for UDDI/Web Services ?
>
> There are some fundamental issues that might prevent this as well as the
political
> ones.
>
> UDDI is a registry only...no repository, where as ebXML has both.  That
could (and
> may need to be) rectified by the UDDI spec (imagine 10,000 WSDL
definitions
> pointed to by a UDDI rep.....how would you manage the storage of all these
in a large
> corporation in a doable fashion?  Put them on different web servers all
over the
> company?  Not likely......some form of centralized repository will
eventually be
> needed to complement UDDI).
>
> The big issue is that UDDI has been designed to be a "global" public
directory
> (though that does not preclude private implementations), with support for
federation
> and distribution of nodes (meaning that the distributed system appears as
a single
> global registry).  Whereas ebXML RegRep has been designed as a
community-level
> registry, to target a specific group of parties with a common interest (a
supply chain,
> industry vertical, etc.) and no provision has been made for
distribution/federation.
> Those are big chasms to cross in trying to merge the two together.
Different
> philosophical roots.
>
> Best practice seems to suggest that you use ebXML RegRep for community
stuff,
> and use UDDI more globally which in turn has a "pointer" to the specific
RegRep.
>
> ...Andrzej
>
> Chaeron Corporation
> http://www.chaeron.com
>
>
>
> ----------------------------------------------------------------
> The ebxml-dev list is sponsored by OASIS.
> To subscribe or unsubscribe from this elist use the subscription
> manager: <http://lists.ebxml.org/ob/adm.pl>
>


----------------------------------------------------------------
The ebxml-dev list is sponsored by OASIS.
To subscribe or unsubscribe from this elist use the subscription
manager: <http://lists.ebxml.org/ob/adm.pl>