Please start any new threads on our new
site at https://forums.sqlteam.com. We've got lots of great SQL Server
experts to answer whatever question you can come up with.
Author |
Topic |
MikeB
Constraint Violating Yak Guru
387 Posts |
Posted - 2003-12-01 : 09:24:32
|
I am developing a multi-user database using Visual C++ / ADO front end and access (soon to be SQL Server) back end. What is the best way of connecting and obtaining recordsets to get the best performance?1) Currently, in my CMyDoc class I have a Database connection object which is opened when the application first begins. I also have a series of Recordsets with the settings : Lock Type = adReadOnly, cursor = adOpenDynamic, cursor location = adUseServer.What can I expect after say 10 users log in simultaneously with the settings shown above?2) Another thought was to create the recordsets as they are required and close the recordsets when finished using them. 3) Using combination of 1 and 2 where the recordsets that are needed frequently would be opened when the application starts and closed when the application ends. 4) When using cursor location adUseServer vs adUseClient, this puts more load onto the .mdb file (server) and less on the client machine, but will this blow up in my face since it is only access that I am using?Any thoughts about the above or additional?Mike B |
|
MikeB
Constraint Violating Yak Guru
387 Posts |
Posted - 2003-12-02 : 07:04:15
|
How do you do it. This thread isn't going anywhere is it?I am quite confused on database design techniques. I know it is best to create one connection object (no confusion here) but it is in recordsets that I am blown away.Some say use server side cursors because it allows real time data and reduces the load on the client maching. Some say use client side cursors because it reduces load on the server, potentially speeds up data access (but can be slower depending on recordset size), and functions such as GetRecordCount() are available. This is just a smidgen of the contradicting opinions in articles that I read. This is what I am thinking.- Using one connection object (obviously)- Using read only server side recordsets for "key" lists. Example the project numbers that are selected to obtain the project information on a form. This would ensure that a project information record added by a user on a different machine would be visible by the user.- Open / Closing recordsets that are for viewing / editing single records as required using SELECT statements, such as when a project number is selected on the project information form. The records, when they can be edited will be opened dynamically on server side so the changes can be seen by other users.This is all I can think of at the moment, been a long weekend and Monday!Any suggestions?Mike B |
 |
|
Stoad
Freaky Yak Linguist
1983 Posts |
Posted - 2003-12-02 : 07:18:48
|
lol Mike.. seems you are pretty good on this subj..What suggestions are you waiting for? :) |
 |
|
MikeB
Constraint Violating Yak Guru
387 Posts |
Posted - 2003-12-02 : 08:16:36
|
Guess I am just under confident, I am self taught so I have to go on instinct alone. I was just looking for responces where someone would outline how he/she approaches the same situation. Mike B |
 |
|
|
|
|
|
|