Object Infrastructures
            for Globally Distributed Persistent Applications

                           Position Paper for
                         the OOPSLA '96 Workshop
                 Object Replication and Mobile Computing

                   Kevin F. Brown and Rajendra K. Raj

                   Information Technology Department
                          Morgan Stanley & Co.
                             750 7th Avenue
                        New York, NY 10019, USA
                          [kevinb,rkr]@ms.com


ABSTRACT
--------

As enterprises continue to become increasingly global and distributed,
they are forced to confront the problems of software application
development in such a setting. One approach is to facilitate the rapid
development of enterprise software applications by developing an
infrastructure or middleware that makes distribution and persistence
completely transparent to application programmers.

At Morgan Stanley, we have pursued such an approach for the past few
years by designing and developing a distributed object infrastructure
for use within the firm for our financial applications. There are
currently over a thousand users of our infrastructure, who are
geographically distributed throughout the Firm's worldwide offices. The
platforms used include various flavors of Unix and Windows, and often
need to integrate with legacy applications on IBM mainframes running
MVS.

Users of our global distributed object infrastructure currently use it
within the context of the Firm's private LANS/WANS. However, with
increased mobility amongst our users, we have begun exploring the impact
of mobile computing on our infrastructure. In short, how do we provide a
truly integrated and distributed infrastructure that allows our
increasingly nomadic traders to "transparently" make use of our services
with integrity and performance.

The issues we have addressed in our infrastructure include object
persistence, transactions at the object level, wide-area replication and
collision handling, security (both authentication and authorization),
support for multi-platforms, multi-languages, and several others. These
are challenging issues in their own right, and require performant
answers in the face of global mobile computing. While we have several
ideas on how to approach our mobile computing, we do not have canned
solutions. Therefore, we look forward to the opportunity to expand on
our experiences and discuss these issues with other workshop
participants.