Database Connection Suggestions

Database Connection Suggestions

Post by timh » Thu, 06 Feb 2003 07:33:22


Everyone, thank you for your help. My first ADO application is almost
complete. Now the time has come to start solving distribution problems.

Version Information:
VC++.NET, msado15.dll, and ADO connecting to MSQL 2000 database

This C++ application connects to a SQL database and modifies the database
using ADO. Once the application is completed, this software will be
distributed to about 1000 clients. Currently, the connection string, user
name, and password are hard coded in the code using the strings generated by
the Data Link Properties dialog box. These connection strings need to be
dynamic in order to support various database configurations. What is the
best way to generate these connection strings? What version(s) of the ADO
should be supported/distributed?

Thank You
Timothy Hughes
First Resort Software


1. Seeking suggestion and advice: how to propagate change from one database to another database


I am looking for suggestions and advice on how to move data from one
oracle database to another on daily basis.

We have a production oracle db (let's call it "prod") on a unix box
that runs 7x24. It supports a web front end application for shopping.
Most of the data in the database will be loaded into memory of an
application server when the application starts.

We also need to provide the capability for people in the shop to change
data in the database. They do not want to change the data in "prod"
instance directly. They want to set up a separate unix box and have a
copy of "prod" running (let's call it "prod_mirror"). They would like
to make changes on "prod_mirror", QA the change and then propagate all
the changes in "prod_mirror" to the "prod" instance. The changes
involve about 10 tables (product, price, catalog, etc). Now I am trying
to find a "good" way to accmpolish this goal. I could think of two
options now:

Option A:
1. Add a flag column in these 10 tables, (possible value: modified,
inserted, etc) to indicate that record either has been updated or
2. Create a db link from "prod_mirror" to "prod"
3. Write an oracle package and move the flagged record in these 10
tables from "prod_mirror" to "prod" when propagating all the changes.
4. Clean up the flag in these 10 tables afterwards.

Option B:
1. Do not change the current table structure. But create a new table
create table DataChangeStamp(
   TableName       varchar2(30),
   ID                       Number,
   ChangeType    varchar2(10),     -- possible value: modified,
inserted, etc
   ChangedBy        varchar2(30),
   ChangeDate     Date)
2. Whenever there is a change in data in "prod_mirror", there will be a
new record created in DataChangeStamp table.
3. Create a db link from "prod_mirror" to "prod"
4. Write an oracle package and move the changed records in these 10
tables from "prod_mirror" to "prod" using the info from DataChangeStamp
when one propagates all the changes.
5. Clean up DataChangeStamp afterwards.

I am leaning toward Option B. What do you think? Any other options?



Sent via
Before you buy.

2. Calender of events

3. Web-Site development/Sybase-HTML Phoenix(Contract)

4. Slow connections over Named Pipes net-lib ... suggestions

5. Pick / Unidata - HIRE ME

6. Connection pooling suggestions

7. Converting datatypes?

8. interface connection suggestions please

9. Any suggestions on how to minimize the number of connections to the server

10. connection string/win98 any suggestions?

11. Wierd connection problem - any suggestions?

12. Database suggestions...