[89] in Open-Software-Foundation-News

home help back first fref pref prev next nref lref last post

OSF Flash - Request for Input

daemon@ATHENA.MIT.EDU (OSF Corporate Communications)
Wed Nov 16 19:00:50 1994

Resent-From: Bill Cattey <wdc@MIT.EDU>
Resent-To: osf-news-mtg@menelaus.LOCAL
To: newsnug@osf.org
Reply-To: newsnug@osf.org
Date: Wed, 16 Nov 1994 16:53:20 -0500
From: OSF Corporate Communications <corpcom@osf.org>

Date:  November 16, 1994

To:    OSF Members

From:  corpcom@osf.org (OSF Corporate Communications)
                       Open Software Foundation

****************************************************

                OSF ELECTRONIC FLASH

****************************************************
An electronic mail news flash for OSF Members from 
the Open Software Foundation
  

                                    November 16, 1994

             Call for Input on Potential
             Transaction Processing PST

A group of OSF members is considering a transaction 
processing PST (TP-PST).  The initiative for this 
potential PST comes indirectly from end-user 
requirements (e.g., SPIRIT).  Briefly, the TP-PST 
provides a communications environment for supporting the 
ACID guarantees (Atomicity, Consistency, Isolation, and 
Durability) of distributed transaction processing -- 
this means either the transaction completes successfully 
or it rolls-back so it appears as though nothing ever 
happened (no inconsistent states or data remain to be 
dealt with manually).

A short description of this potential PST is attached.  
The PST comprises a XAP-TP stack which provides 
interoperability across the Internet.  However, it will 
also run over virtually any connection-oriented network 
(e.g., SNA, DECnet or OSI).  The PST provides the 
middleware needed by the industry to support TP 
monitors, distributed data bases, and the X/Open TP 
interfaces (XATMI, Open-CPI-C, and TxRPC -- see the 
attachment for how the XAP-TP stack fits with the X/Open 
Model).  All components of the PST are implementations 
of X/Open, ISO or Internet standards.

The reasons for doing this PST are

a) accelerate the widespread use of standardized 
   distributed transaction processing (DTP) by putting a 
   common DTP reference implementation into the open 
   arena, i.e., make DTP ubiquitous,
b) promote interoperability through the existence of a 
   common reference implementation,
c) provide a foundation for future PSTs implementing 
   X/Open's standardized TP interfaces,
d) reduce the cost of this common foundation by sharing 
   the cost of testing and maintaining this reference
   implementation among the PST sponsors,
e) establish a base for future shared implementation of 
   the TP Commit Optimizations and Subtransactions.

The purpose of this letter is to request feedback to 
determine whether the OSF membership believes this is a 
worthwhile PST and how many vendor and/or ISV members 
would be interested in licensing this technology if it 
were to be made available in a reasonable time frame.  
The group considering this PST are not looking for a 
hard commitment at this time.  However, a meaningful 
indication of how many potential licensees might be 
expected is essential to deciding whether the group will 
go ahead with the PST or not. As the group would like to 
make a decision before the end of the year, please 
respond to the following question by the end of 
November:

     If there is a TP PST, would your company be
     likely to license this technology?


Send responses or requests for further information to:

Henry Lowe
Open Software Foundation            Tel: +1-617-621-8783
11 Cambridge Center                 FAX: +1-617-621-7324
Cambridge, MA 02142                 E-mail: lowe@osf.org
USA
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

                Draft Proposal for a TP PST
                 V 3.0, 12 October 1994


This is a brief description of a PST intended to bring 
Transaction Processing (TP) into the OSF set of 
offerings.  The PST consists of:

  a) X/Open's XAP-TP interface;
  b) an OSI TP stack (with the commit functional unit); 
     and
  c) an RFC 1006 mapping to the Internet.

Figure 1 below shows this interface and stack in the 
context of the X/Open TP model.

                   +----------------+
                   |                |
                   |        AP      |
                   |                |
                   +----------------+
                   /       |        \
                  /        |         \
                 /         |          \
      E.g., SQL /          | TX        \E.g.,TxRPC (IDL)
+--------------+    +-------------+     +--------------+
|              |    |             |     |              |    
|      RM      |----|     TM      |-----|       CRM    |
|              | XA |             | XA+ |     (U-ASE)  |
+--------------+    +-------------+     +--------------+
                                                 |
                                                 |
                                          XAP-TP |
                                  +--------------------+
                                  |  Functionality of  |
                                  | OSI TP with Commit,|
                                  | OSI Upper Layers   |
                                  | and RFC 1006       |
                                  +--------------------+
                                             TCP |
                                                 |
                                          Connection
                                           Oriented
                                            Network
                       Figure 1


This PST is for middleware, i.e., it does not interface 
to users' TP programs (the box labeled AP in the 
figure).  The PST provides a standardized interface 
(XAP-TP) for TP Monitors and will operate over any
connection oriented (CO) network (e.g., TCP/IP networks) 
offering the usual CO guarantees.  The TP stack is based 
on ISO standards with the single exception of RFC 1006 
which is an Internet standard.  Thus, the PST provides 
interoperable, portable, standards based TP 
functionality.  The base technology for this PST exists 
today.

While not being proposed at the moment, if this TP PST 
goes forward, this PST would from an excellent basis for 
other PSTs for TP technology (i.e.,X/Open's three TP 
interfaces and further work in this area such as 
Queuing).












home help back first fref pref prev next nref lref last post