Gary Nolan wrote:
> Please excuse me if this question has already been answered.
> What am I going to gain/loose if I use B-trees Vs Superdex level 1.
>

Well, going to b-trees from Superdex L1 you will lose at least:

1) concatenated indexes
2) subitem indexes
3) Item truncation
4) Indexes on manual master sets
5) >16 indexes per dataset
6) Possibly grouped indexes (I don't remember if this was part of L1 or L2)

Some basic design issues:

Another thing you might have to worry about is the number of datasets allowed
in a database.  Using Superdex, you use 1 to 4 (depending primarily on your
choice, unless you have a very large database) datasets for indexing, which
leaves you 195 to 198 datasets for data.  With Image b-trees, you will use one
dataset for an index, unless the index is used in more than one set.  For
example, if you index Customer-Name, and this item is used in two datasets, you
can index both using only one automatic dataset.  If you want to index
Customer-Name and Product-Number from one set only, you would use two automatic
datasets.  Each detail dataset can only have up to 16 indexes, and each
automatic master dataset can have up to 16 details linked to it.  If you wanted
to index Product-Number in 20 detail datasets, you could do this, but you would
need two automatic master datasets.

What you gain with b-trees:

1) Support costs included with support for Image
2) If you don't already own Superdex or Omnidex, you don't have to buy
   one

There are arguments about which of the three (b-trees, Superdex or Omnidex) has
the highest performance.  I don't know of anyone who has actually done a
controlled performance test on all three.  My experience, and therefore my
belief, is that each one has the "highest performance", depending on what you
are doing.  Which one performs best overall -- again, it depends on what your
overall application is doing.

There are also some arguments about which one of the three has the simplest
design (easier to understand the internals).  I won't argue this because I
think everyone's previous experience affects your ability to understand a
design quickly or at all.

There are probably a couple of items I left off of both the gains and loses,
but this gives you some idea.  Generally, Superdex L1 and Omnidex IMSAM
products have more features and accept more indexes than b-trees, but you buy
b-trees when you buy Image and your support cost is one line item.  Each has
its advantages and disadvantages.

IMHO,

LB