Re: Comments requested on VB4 OLE2 client/server architecture with multiple RDBMS

From: Derek Cummings <derek_at_dac1.demon.co.uk>
Date: 1996/02/06
Message-ID: <823569546.23426_at_dac1.demon.co.uk>#1/1


"P. Jespersen" <pjespe_at_hevanet.com> wrote:

>What you are about to do sounds, from an architecture standpoint, almost
>exactly like what we are in the process of finishing, but with one big
>difference. We are only using OLE2 on the client side. We brought in
>Microsoft consultants, and their opinion was emphatically that DCOM and
>Distributed OLE2 is NOT ready for prime time in a multi-threaded,
>thread-safe server. They indicated that not only was it NOT stable,
>but that the performance would be a real dog. They would not even
>consider it at this time.

Our architecture does not prescribe the use of distributed OLE2, in fact I agree with you completely that at this stage DCOM does not sound like it will be up to the job. We believe our architectural approach should allow us to distribute selected layers within our application when DCOM eventually becomes a viable option.

We will be using OLE2 on the client side only. The logical Data Access layer will probably be composed of components which sit either side of the client/server divide on the network. We are now discussing the use of stored procedures within the RDBMS for performance reasons i.e. the Data Access layer would consist of an OLE2 object on the client PC exposing properties/methods where the methods would call stored procedures across the network. An alternate implementation of the same Data Access layer could expose the same properties/methods but utilise ANSI standard SQL queries to interact with the database.

---
Derek Cummings             
Leeds, UK

EMail:  derek_at_dac1.demon.co.uk
Received on Tue Feb 06 1996 - 00:00:00 CET

Original text of this message