Rudi wrote
 
As we realized in our department there is a problem with a master-dataset :
More than 20% of all entries belong to entries where the hashing-algorithm had
to calculate a secondary key-entry. Due to that the performance is very bad.
The master-dataset is 50% full. Is there a way to optimize the dataset-filling
? How does the hashing-algorithm work ? (I know there must be an article
describing it )
 
Leon writes:
 
The hashing algorithm is described in the image handbook, a 10+ year old
robelle, adager & friends production which is fun to read.
As you will no doubt know the number of secondaries in a masterdataset is determined
by the way your key is constructed and the capacity of the set.
 
Some people (among which our adager friends & hp) say taking a prime for the capacity
is a good rule, but I was a bit more stubborn than that and wrote a small programme
that takes an exsisting dataset and computes the number of secondaries at a range of
capacities. I have reached amazing results with it. You'll find that there are almost always
lower capacities which give less secondaries and boost up performance.
I'll dig up the source tomorrow and let know where you can get it.
It's cobol and has been tested on datasets up to 1 million
records.
 
Another approach is optimizing the key. A J? always better than an X?
The first 4 positions in a x? decide mainly on the hashing, so it may be worth while to change
the order of data in the key. If most keys start with something identical.
I have seen some cases where people used an x? to store a numeric key. This is very unwise.
 
Frankly I thought no-one cared anymore for this sort of tuning stuff. Buy more
Iron(by which he means hardware) my manager always says when we bring up the tuning subject.
 
Leon Degeling
VrijUit bv.
 
I say what I like, my boss probably disagrees.