I was woundering if a simple perform find or a matching related record would be the faster way to "search" a table with 100,000 records or more.

Is it a question of number of search criterias? Does it depend on the number of records?

The search criterias would be exact matches or ranges, which is easily implemented into FM7 relations.

/Johan

OK, I think I finally got an answer to this myself.

I made some tests with about 150,000 records with a shared DB via FMS, just to make times a bit more measureable.

These are my results:

If the criterias are exact matches only, then there are no measurable differences. In both cases the search took about a second (or less).

When it came to ranges though, there was quite a big difference. I made up a relationship with three exact matching fields and a date range. This relation match took about 6 seconds, while the ordinary search (with the exact same criterias) was about as fast as before!

Conclution:

Exact matching searches can be made thrue a relationship just aswell as thrue an ordinary search.

But as soon as it comes to more difficult relation matches such as ranges (or just a "less than") the relationship is outtimed by the ordinary search by many times.

I hope someone else can have any benefits from this!