This article is part of the series An Exercise in Performance Tuning in C#.Net.
At this point I did not re-run the profiler but continued investigating memory allocation. Here's an interesting looking result:
(removed link to missing image file)
in Visual Studio's debugger, you've landed on an Exception statement. You look at the stack trace but it just points back to a custom exception class you've created. But you know that code is good, it can't be throwing the error. What's going on here?
this is an easy mistake to make. I've run it across it in code from a number of people, and recall making the mistake myself at one point. The problem is most likely due to nested throws to a new custom exception.
This is not part of my series on performance tuning a specific app.
I have a Registry class in which I want to place a generic collection of objects. In this way I can add new items to the registry on the fly (i.e. from a user application), without having to recompile the library containing the the class. What is my best option, in terms of performance, for a .Net 2.0 collection to hold my mixed-bag of objects? I'll be referring to these with a string name.
This article is part of the series An Exercise in Performance Tuning in C#.Net.
Where else could I improve performance? I thought I should inspect the memory usage and garbage collection. For that I found a great little tool from Microsoft, the CLR Profiler, which I found through the MSDN Patterns & Practices series of guides on application performance and scalability. The specific article that I found most helpful was How To: Use CLR Profiler.
This article is part of the series An Exercise in Performance Tuning in C#.Net.
The application I was working on makes a number of calls to a remote server to perform operations. These calls are made through a proprietary API implemented in .Net. Speaking with the vendor, I discovered two really dumb mistakes that were killing our performance.
Situation: We have a stored procedure running a query whose WHERE clause is
given as a parameter. No, that's not the problem in and of itself, at least not
today. (Treat this as a non-negotiable requirement for now). Within that
WHERE clause there might be a query against a field, call it myField. This field
is a varchar and wildcards are not used. Platform: SQL Server 2005.
myField has been changed to a varbinary field and holds an encrypted value — thus can no longer query directly against it. How do we make this work?
Your application is slow. Horribly slow. And that just isn't acceptable.
You've transferred or run a bunch of stored procedure scripts, but you can't execute them. Reason - execute permission denied. You forgot to put a grant statement in your script.
The trivial solution is, of course, GRANT EXECUTE ON {your proc name} TO PUBLIC. Slightly less trivial is to grant to a specific role, but most
people needing this tip will only be using PUBLIC.
Wouldn't it be great to automate this for all stored procedures in the database? Well, here you go:
On a Microsoft SQL Server 2005 installation with a linked server
configured to "myserver" (which happens to be 2000), execution of a remote
stored procedure (EXEC myserver.mydatabase.dbo.mysproc) fails with error:
Msg 7411, Level 16, State 1, Line 1
Server 'myserver' is not configured for RPC.
Microsoft .Net Framework 1.1 and .Net Framework 2.0 don't play well together (as ASP.Net apps) on Windows 2003 Server 64 Bit Edition.
either upgrade any ASP.Net 1.1 to 2.0 or switch to 32 bit compatibility mode. Of course switching to 32 bit mode will cause you to lose out on some performance benefits, but maybe you're okay with that (unless you're running under a very high load). Two easy steps:
