Separating database system tables from the rest.
Posted in 2000
Topics: Storage & Space Management, Server Administration, Migration, Import/Export & Data Conversion
One of the issues I've had to deal with is the fragmentation of the
system tables within my larger databases. I thought that separating
those tables from all of the user tables within a database might be a
great ideas, for example:
create database asdfg in asdfg_system . . . . ;
I guess my thinking would be:
1. Any growth of the system tables wouldn't affect anything else except
system tables.
2. I would be able to reorg at the dbspace level for ALL dbspaces, as I
could unload tables, drop and recreate dbspaces, and recreate and reload
tables without bypassing those dbspaces containing system tables.
Any thoughts? comments?
--
John Carlson
Informix DBA
WHSmith USA
#include std_disclaimer.h /* These are my opinions, not my company's
opinion */
"Carlson@WHSmith" wrote:
>
> One of the issues I've had to deal with is the fragmentation of the
> system tables within my larger databases. I thought that separating
> those tables from all of the user tables within a database might be a
> great ideas, for example:
>
> create database asdfg in asdfg_system . . . . ;>
> I guess my thinking would be:
>
> 1. Any growth of the system tables wouldn't affect anything else except
> system tables.
> 2. I would be able to reorg at the dbspace level for ALL dbspaces, as I
> could unload tables, drop and recreate dbspaces, and recreate and reload
> tables without bypassing those dbspaces containing system tables.
>
> Any thoughts? comments?
Yes. Don't worry about the fragmented system tables. In OL5.xx this was
a major problem. In IDS 7/8/9 it is NOT a problem if your Data
Dictionary Cache table is large enough to hold the dictionary entries for
all of your active tables. You can expand the cache using the
undocumented, but supported, ONCONFIG parameters DD_HASHMAX and
DD_HASHSIZE. The former is the number of entries per hash bucket and
defaults to 10, the latter is the number of buckets and defaults to
either 31 or 128 depending on version (see onstat -g dic output).
Since IDS now caches the data dictionaries for the most active tables and
the cache size of adjustable, fragmentation of the system catalog tables
is no longer an issue, except as it may affect the fragmentation of data
tables. And of course you can always reorg the fragmented data tables.
Art S. Kagel
Related threads
- Posting from the Informix-list
- Migrating from IDS 9.40.UC6 to 11.50.UC3
- Ip for a network session
- questions onstat -g