Comments on: Troubleshooting SQL index performance on varchar columns

By: Marc Scheuner

Marc Scheuner — Wed, 06 Mar 2013 10:09:00 +0000

Interestingly, with Entity Framework (an ORM), you *can* define whether or not to use Unicode strings – with things like data annotations on your model classes, or with .AsNonUnicode(). On the other hand, with *RAW* ADO.NET (which certainly doesn’t “hide” the database in any way, shape or form), you can’t seem to define this – it will always use N’…..’ for your search parameters – even if the underlying column is a “varchar” – thus killing your query performance …..

By: Erik

Erik — Thu, 31 Jan 2013 09:30:00 +0000

Excellent! Just sent it over to our developers

By: Simon Skov Boisen

Simon Skov Boisen — Fri, 03 Aug 2012 06:18:00 +0000

Had a similar experience the other day where I were accidentally comparing a string with an integer (i simply forgot to put ticks around the number when I did the query) which meant that for every single row MSSQL were converting the string to a integer. I looked at the query plans, everything looked as it should – that really made me scratch my head until one of the
database gurus in my company helped me find the error simple as it were.

By: James Banner

James Banner — Fri, 20 Jul 2012 05:43:00 +0000

Awesome post, I didn’t knew it, learned something new.

Just goes to show the software engineer’s dream to “hide” complexities or to make “persistence ignorance” as far as database goes is just a pipe dream which we can live in, until a problem comes knocking.

By: Michael K. Campell

Michael K. Campell — Thu, 19 Jul 2012 20:55:00 +0000

Jimmy, this is a text-book example of a problem that DBAs and perf gurus typically call data-type coercion – where data-type mismatches (as you clearly outlined) invalidate the use of an index.

And, as an FYI, you can EVEN run into this in situations where the column is, say, nvarchar(20), and @p1 is defined as nvarchar(10) – because you’ll run into the same problem with coercion.

By: Anonymous

Anonymous — Thu, 19 Jul 2012 13:23:00 +0000

Ha! What’s the alternative?

By: Andrei Rinea

Andrei Rinea — Thu, 19 Jul 2012 13:08:00 +0000

That’s why I hate ORMs. They solve n problems and create n+1.

By: My Java Coding Still Has A Problem? | cheap airfare flights, cheap airline tickets, cheap package vacation deals

Thu, 19 Jul 2012 09:59:16 +0000

[...] bounds. You gotta a least post relevant code to get help! Give your answer to this question below! Question by : My Java Coding Still Has A Problem? import java.util.Scanner ; public class Flight_Ti…H, F, T, Q, Type, Z; System.out.print("ttttWelcome To ZulDin Flight"); [...]

By: The Morning Brew - Chris Alcock » The Morning Brew #1149

The Morning Brew - Chris Alcock » The Morning Brew #1149 — Thu, 19 Jul 2012 08:29:14 +0000

[...] Troubleshooting SQL index performance on varchar columns – Jimmy Bogard discusses an interesting performance problem he encountered with indexes and the use of an ORM caused by a difference in data type (ANSI vs Unicode). Definatly one to be aware of. [...]

By: Larry Silverman

Larry Silverman — Thu, 19 Jul 2012 03:51:00 +0000

Hi, Jimmy. Are you using Enterprise Edition? If so, you can enable data compression and use Unicode for simplicity and consistency everywhere.