Plateforme Level Extreme
Abonnement
Profil corporatif
Produits & Services
Support
Légal
English
Best? way to process a huge result set over a network
Message
De
01/12/2004 01:48:11
 
 
À
30/11/2004 16:39:36
Information générale
Forum:
ASP.NET
Catégorie:
Autre
Versions des environnements
Environment:
C# 1.1
Database:
MS SQL Server
Divers
Thread ID:
00965818
Message ID:
00965975
Vues:
10
>I assume you mean Visual FoxPro. I can only consider C# solutions. In fact, my company is switching from VFP to C# largely to address performance problems between VFP and MS SQL 2000. I assume that there is a optimal method in C#.

This is NOT about languages, but about how/where to process the data. The paradigma of [ado].net is to work on rather small sets in [client] memory. Yours is one of tasks where working on the client disk seems best: the workload is offloaded from the server and the amount of data seems to be to large for memory processing, forcing swaps [first assumption on the reason of slowness, should be verified. Second assumption: running the report preparation on the server is ill advised, because it either blocks other requests or takes too long - also to be verified in advance, but based on your description<g>].

Feeding the data into a local instance of MSDE/SQL server would be a c# solution, as would be a solution using any another "local data handling package" for .net. Please google for yourself - besides vistaDB and codebase [I haven't used both] there are quite a few others. I guess deployment needs/licening and dev training cost would be the most influential parameters to check.

It is exactly the area where Visual FoxPro works best and calling a vfp COM object in a .net solution for this special case might even be the most cost effective solution because some vfp developer knowledge exists at your place. But the other alternatives should be not far behind in performance, if it gets "political". If running those reports on handhelds is deemed necessary, you also have a technical reason *and* technical filtering criteria for local data handling packages<g>. If you decide to work with such a package, pls. keep me informed on the packages evaluated, choosen and on your impression after some expirience. On the other hand: 300000 records should keep you well below MSDE's 2 gig barrier, and processing speed of MSDE and SQL server is identical (on the same machine), if there is only one concurrent query set running. If you have to run the report only on one workstation, either install MSDE or one of the cheaper SQL server licenses. If you have to run the report on many workstations, I'ld lean because of security and install issues to a dedicated package [including vfp <g>].

my 0.02 EUR

thomas
Précédent
Suivant
Répondre
Fil
Voir

Click here to load this message in the networking platform