Level Extreme platform
Subscription
Corporate profile
Products & Services
Support
Legal
Français
SQL Server slower than VFP?
Message
From
08/08/1999 20:36:34
Charlie Schreiner
Myers and Stauffer Consulting
Topeka, Kansas, United States
 
General information
Forum:
Visual FoxPro
Category:
Client/server
Miscellaneous
Thread ID:
00250319
Message ID:
00251317
Views:
16
Mike,
It's one thing to store the data in one location, it's quite another to process the query in only one location. I am only interested in multi-user apps, as perhaps are you.
My point was that in a file server system, the file server doesn't do much. Each user's CPU is were all the work gets done. If there are bandwidth considerations, then that may not be the best place, but if bandwidth isn't a problem, having a CPU totally dedicated to me and my work is pretty cool. With SQL Server, my SQL statement gets processed sharing the CPU(s) with everyone else doing a query. The more users on the system, the less CPU time I get for my query. With VFP in file server mode, I get my own CPU.
Any aggregate functions where a large data set is required, SQL Server has the records and doesn't have to transmit them, but if the data can be accessed via indexes and the resulting record set is relatively small (few 1000), I would think VFP-File Server would be faster on the graph, the more users you add. I realize this is highly conditional. Having 100 CPUs vs. 1 was made years ago by George Goley and I was just remembering his point.

>>>If you were to graph performance, SQL Server would start out with a higher response time for one user, but would not grow as step as VFP as more concurrent users are added. The real challenge is determining at which point VFP or SQL makes sense in your environment. Not to mention remote users or internet users of you product.
>>>
>>Mike,
>>I have no experience with SQL Server, but I wonder about your logic. With VFP, when I add a user, I get a new CPU to do the work. With SQL Server, I must share the CPU(s) in the server with everyone else.
>>Charlie
>
>My assumption is that you plan to write an application that is used by more that one person. Therefore the databases must be stored in one location. If you are writing only a single-user application, why would you even consider SQL Server?
>
>My point is that benchmarking a query in the environment that you previously stated is not a valid test for a multi-user application. VFP will naturally be faster working with smaller datasets and smaller number of CONCURRENT users. This is what it is designed for. After a point as the load / dataset sizes / contention increases, SQL server will handle the load better. This is what it is designed for.
>
>Mike
Charlie
Previous
Reply
Map
View

Click here to load this message in the networking platform