Article View: borland.public.cppbuilder.database.multi-tier
Article #1699Re: Making Midas useful
From: Michael A. Feinstein
Date: Fri, 06 Aug 1999 00:00
Date: Fri, 06 Aug 1999 00:00
51 lines
2810 bytes
2810 bytes
That is the default behavior. To correct it , design you own TDataset descendant which would store results in memory and write them back only when needed. That is what I did and it proved to be successful. The only drawback is that you do not see changes done by others to the database (only by the application using your dataset). e-mail me if you want more details. peterayre@enterprise.net wrote: > > Please feel free to correct me or point out any mistakes in what I am about > to post. > I have not been using the Midas stuff until now and admit that I may be > missing something. > > I have a multi-user system around 30 on our larger set-up. > This is attached to SQL Server using ODBC. > > These machines are all querying away in a time critical application and > return data that is so similar that only filters would be required to > differentiate between any two set-ups. > > currently this means that from 30 different sources a single query comes in > about twice a second. > This is a considerable load for a complex query on a large and busy table. > > I had an Idea and thought that if I use the Midas components then I can make > this query once and then distribute it to many machines. > This seemed like a fine use of the multi tiered functions. > > however a little checking shows me that when I try this I am not simply > returning the results of a single query to whichever applications request > them. I am re-querying every time. Using SQL Trace shows me the same query > hitting the server very time I open a new Client or refresh the view in an > existing client. > > I want to make Midas work for me but if this is the normal behaviour then > what is the point. I cannot see that it makes any difference if I query > directly from 30 points or get a single point to query for me 30 times. > I would be better setting up a temporary paradox table on a single machine. > > It may be that I could loose the ODBC connection under this regime but on an > NT based system I cannot see any gain. > > As I said at the top I could be missing the point or just plain wrong but > PLEASE could someone enlighten me here. > > Thanks > > Peter
Technical Details
Message-ID:
<37AB107F.8B4B7045@san.rr.com>Path:
arc1-ca.pugleaf.net!unknown!not-for-mailReferences:
<7o4bqc$k126@forums.borland.com>Show Full Headers
Complete Article Headers
X-Google-Language: ENGLISH,ASCII-7-bit X-Google-Thread: fcd2b,5201f9e5da3254 X-Google-Attributes: gidfcd2b,public From: "Michael A. Feinstein" <mfs@san.rr.com> Subject: Re: Making Midas useful Date: 1999/08/06 Message-ID: <37AB107F.8B4B7045@san.rr.com> X-Deja-AN: 509642858 Content-Transfer-Encoding: 7bit References: <7o4bqc$k126@forums.borland.com> X-Accept-Language: en Content-Type: text/plain; charset=koi8-r Organization: Another Netscape Collabra Server User Mime-Version: 1.0 Reply-To: mfs@san.rr.com Newsgroups: borland.public.cppbuilder.database.multi-tier