Hi All,

I was just thinking about the way Sybase stores the data and am
wondering what will happen if it is stored based on value. That is,
create an extent for each character in the character set for each
column. Think of it like a two dimensional array with 'columns' are rows
and each character is a column in the array.

Characters ........
columns
.
.
.

Each extent grows based on the character for each column. Data can be
inserted at
the end and sorted by a system process.

Since the essence of DBMS is storing related data and retrieving it,
wouldn't it be
faster this way? I know the optimizer need to work entirely different.
But just curious to discuss about the pros and cons of this.

My second thought is integrating Sybase to the OS strongly. That is,
Sybase ASE is not just a CD. The machine itself is configured and ready
Sybase. Do you see this or any of it's variations in future?

Thanks

Ganesh.