Wow, I didn't realize this was such a deep subject! All of you have certainly made some interesting and valid points.
I do understand that having many indexes will slow down writes (but this is not a real noticeable thing in our apps here and has never appeared to be a problem even though we have many tables with half a million and more records and many indexes on those tables). And I also undertand what you mean by the fact that good coding is needed when building the SQL Selects to fully use the power of rushmore. All of those things are good points and very true.
But my question is JUST THIS. Do more indexes on one table IN AND OF ITSELF really slow down a SELECT??
Charlie, you did say that...
>> If you have expressions that don't do a good job of specifying a small set of
>> records, you might read 20 megabytes of index info over the network. See
>>
http://www.networksplus.net/cschreiner/OutFoxed15.html for an explanation as
>> well as Chistof Lange's message, Thread #
257424 Message #
259556.
This statement does confuse me a little. I read the articles you mentioned but I don't see how 20 megs of data has to be moved over a network because the index file has got a lot of tags in it.
Surely you don't mean that the WHOLE .CDX file is copied somewhere EVERYTIME you do a query? Are you talking about just the index expressions or do you mean all the data for the fields in the WHERE clause of the query??
If you just mean the index expressions then I have a hard time believing that an index file would contain 20 megs of index expressions.
What I am assuming you mean, and yes I hesistant to use that word, is that all the data in the CDX file that falls under the index expressions that would be used for rushmore in the SQL would be copied over somewhere. If THAT is true then I don't see how that would make any difference whether I had ONE tag or ONE HUNDRED tags since it would ONLY copy over the ONE tag anyway if just one field was in my where clause??
Forgive me for being slow but I still just don't see how more tags in and of themselves would slow down a SQL, unless when you say, "...you might read 20 megabytes of index info over the network...", you mean that it reads THE WHOLE INDEX FILE regardless of which indexes are used. If that is true then I could see it.
Maybe I'm over analyzing this... especially for a FRIDAY AFTERNOON!
