TMG-L Archives

Archiver > TMG > 2004-10 > 1096762314


From: "John Cardinal" <>
Subject: RE: [TMG] Splitting a V5 Project
Date: Sat, 2 Oct 2004 20:11:54 -0400
In-Reply-To: <6.1.2.0.0.20041002190007.03358eb0@incoming.verizon.net>


Walt wrote:
> At 10:27 AM 2/10/2004, Lee Hoffman wrote:
> ... certain filters will be faster than others, but even so they are still
> searching the entire project.
>
>
> Lee
>
> Are you still searching the whole project if you deactivate data sets?

Walt,

I'm not Lee, but I'll jump in. In TMG v5, there is only one database per project. All the datasets for a project are stored in that one database. At a finer level of detail, all the name records are in one table, <projectname>_n.dbf. Even if the dataset is "disabled", the program still have to load some of those records as it looks for names that are part of the disabled dataset. Indexes and other factors help TMG avoid looking at data it doesn't need, but some overhead just can't be avoided.

In the end, you can form your own opinion. Close all applications except TMG. Make a big project, possibly by adding multiple datasets. Run some reports that use filter on a single dataset. Note how much time the filters take to process. Then disable all the datasets but the one used by the filter in the report. Rerun the reports. Any faster?

Please note that a small difference doesn't mean anything; even when no applications are running, your PC is running background tasks and a small difference could be due to some of that activity. To minimize that effect, you should perform the two tests (datasets not disabled, datasets disabled) multiple times and use the average. That's still only an approximate comparison, but it will be good enough for these purposes.

You may not care enough to do the comparison, but if you really want the answer, doing performance tests is the only way.

John


This thread: