XPost: uk.politics.misc, uk.gov.local, uk.gov.social-security   
   XPost: alt.politics.british, uk.community.social-housing, uk.com   
   unity.voluntary   
   From: @foster.com   
      
   A Distributed Objects Overview   
   A study of the DCOM system that Microsoft is developing Waite Groups p. 323   
   reveals the problems that Microsoft is trying to solve.  A distributed   
   system is many things to many people so far as the definition is concerned.   
   Waite Group's book states their definition of a distributed system is p.   
   323, "a collection of software entities physically spread out across two or   
   more computers, working together to achieve a goal in common".  The   
   definition here includes a file sharing arrangement, but some computer   
   programmers would not define a distributed system as that.  Yet the Waite   
   Groups says that whether the computers are aware that files exist on other   
   computers, not locally, does not matter at all.   
      
   There are however, those distributed systems built with explicit knowledge   
   of a computer network or communications link.  These systems are designed   
   from the beginning so that the system's constituent software components are   
   executed on separate machines  (an object-oriented approach).   
      
   DCOM is the outgrowth of the evolving technology called distributed objects.   
      
   Initial client server implementations were designed to place some core   
   application functions such as user interfaces, and simple data validation   
   routines locally, at the client user's machine.  And an interesting   
   consequence of client/server computing is that the mainframe can be turned   
   into a sort of super-server, providing database and business logic services   
   for all of the PC based clients.   
      
   The mainframe server is not too far back in its time of usefulness though.   
   A mainframe is the embodiment Waite's Group p. 324 of centralized computing,   
   and it is an archetypal example of a nondistributed system.  The need for   
   mainframes has actually increased in spite of the option for distributed   
   systems.  Credit card companies, and insurance firms, and telephone   
   companies (late 1990's) possess a massive volume of information processsing   
   need.  Only mainframes can be used.   
      
   Two, and Three-tier Network Architecture   
   Client/server computing has been evolving.  The two-tier model is based upon   
   a client having all the user interface functionality, and the business logic   
   required by the system (fat client).  While the server is the second-tier   
   where the application data is located.  Application data is often stored in   
   some type of a database and is often stewarded by a database server program.   
      
   The client applications which may be many in number, each running at its own   
   client workstation -send requests to the database whenever the application   
   needed information or data stored in the database (i.e., inventory   
   database).  A two-tier client/server system is still a good fit for a   
   smaller departmental type of application.   
      
   However, the two-tier architecture does suffer from several deficiencies.   
   For one, each client requires the working interface with much of the   
   business rules portion of the application.  So the application as a   
   multi-user system is dependent on each client, and the clients can be too   
   many to service at the level needed for tweaking of the applicaton.   
      
   In addition to the upgrade/maintenance issues above, the database itself is   
   connected to all the clients and changing the structure of the database in a   
   significant way will also affect the ability of the many clients which   
   search the server database by direct connection with its DBMS.  And it is   
   also more difficult to use more than one application running on/per client   
   when the data for several applications all must agree to the same DBMS   
   format.  And a change in any application on the clients could again affect   
   the multi-user system of clients to the degree that their performance of any   
   other tasks may be stopped by a need to repair and upgrade that one   
   application being altered.   
      
   One can yet imagine the solution to the problem issues of a two-tier network   
   architecture.  The issue of several applications running on a single client   
   being dependent on the single machine, can be addressed by giving each of   
   the applications an application server.  A corporate research department has   
   a server, and the financial department has a server, and personnel has a   
   server for its application also.  But unless we are considering an   
   enterprise with 5000 employees, too many servers can be a problem.   
      
   Three-tier Network Architecture   
   But first consideration is that you do not want a system where all  the   
   clients are directly connected to the main server.  So it is necessary to   
   have at least two servers, which makes this a three-tier network   
   architecture.  But the use of object oriented-programming can allow more   
   reliability of use for an application server which has been (would be)   
   designated to run several different applications.   
      
   In a three-tier architecture network system the clients still maintain   
   responsibility for the user interface, and perhaps a limited amount of   
   validation.  Because these clients in a three-tier architecture contain a   
   minimum amount of functionality, these programs are sometimes referred to as   
   thin-clients.  The use of the web itself to manage individual clients, i.e.,   
   the Microsoft Windows operating system updates is not related to the core   
   functions of the application.   
      
   There is some basis to the fact that the use of a good relational database   
   which is object-oriented can be shared between applications.  So that part   
   of the server issue can be partly controlled.  The remainder is probably   
   related to sufficient programming of applications and the use of classes   
   that will work well together.  The  application suites such as Microsoft   
   Office and Lotus Smart Suite can be considered as classes which can be   
   managed using script.  And most of this application becomes more manageable   
   from the use of such integrated applications as a foundation.  But what I   
   really wanted to mention was that early techniques for using multiple file   
   compiling and #include files, in C language were a method to change parts of   
   the program without interfering with the interface, or some of the   
   non-involved objects.   
      
   Systems and Models   
   A system is an assemblage of interrelated elements.  One example is a   
   mechanism, which may mean the related parts of a machine (as in computer   
   processor, memory, and other components of a computer) or the related parts   
   or stages of any process (the mechanism of creation in art and science, the   
      
   [continued in next message]   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)