anupkumar
6th March 2002, 12:08
Hi markus,

can you recall or do you have the document in which the usage of the storage2 file is mentioned.

victor_cleto
6th March 2002, 12:25
Splitting of rdbms storage files in Baan4c4 for achieving a better performance in the database driver startup
Author G Thomes Creation Date: 21 Jan 2002 Alternate ID:
Solution No: 121726 Solution Version: Status: Published
Product: port6.1c.06 Sub Product: Session: storage files
Package: tt Version: B40 Release: c
Solution Type:

Symptom information
slow session startup after RAM upgrade

Solution description
VERSION(S):

Portingset form 6.1c.x onwards

SITUATION IDENTIFIED IN:
Baan4

ATTACHMENTS:
n/a

SITUATION DESCRIPTION:
The parsing of the <rdbms>_storage files takes a lot of time when a lot of lines are defined.
As the functions that handle the <rdbms>_storage file use an algoritm of order n-square, these lines functions cause slow driver startup at sites that have specified a lot of lines in these files

SOLUTION DESCRIPTION:
For BaanIV there is only the <rdbms>_storage file. (So no _driver_param and _storage_param file).

The porting set is also looking at <rdbms>_storage2 ( Not documented and not supported by sessions).
This file has the same layout as <rdbms>_storage.

Here is a simple program flow on how the driver reads the files <rdbms>_storage and <rdbms>_storage2:

IF <rdbms>_storage2 EXISTS
THEN
idxopt, refresh time and private/group are read from <rdbms>_storage2. The storage part in this file is not used and can be empty.
As these parameters are often the same for all tables, this file is very short. For Oracle level-2:
*:*:T:group:0214:5:

The storage file is alway read from the <rdbms>_storage file.
As the storage clause can differ per table, this file can be big. Please note that the fields idxopt, private/group and refresh time are ignored.

This file is only accessed during creation of tables (when the storage clause is needed).

ELSE
Everything is read from <rdbms>_storage
FI

EXAMPLE(S):

WORK-AROUND:
A frequently used work around: use the original file when creating tables, and replace it by a stripped one (just the default line) when the end users are doing OLTP.
This works as the storage spec lines are only needed during table creation.