Hello, Tomasz:
There is not a simple answer to your question.
OS/400 and i5/OS (or IBM i) is primarily a paging system. Real main 
storage is a "cache" for objects residing in single level storage (on 
disk). So, if the real main storage is large enough, once objects get 
paged in, they will tend to remain in main storage, thus increasing the 
speed of subsequent accesses. For files and tables, this will not help 
much on the first access, but could help speed up subsequent accesses. 
The larger the cache, the more objects (and parts of objects) can reside 
in main storage for fastest access. So, adding more main storage to an 
i5/OS system or LPAR will not hurt and usually helps.
You can also use techniques like SETOBJACC to tell the OS to load 
certain frequently used objects into a designated main storage pool.
But, you mentioned trying to improve the time for SQL queries to return 
results, so adding more main storage might not be the first thing to try.
Are you using the available tools for tuning your database, such as the 
Index Advisor in iSeries Navigator, and running jobs in debug to see the 
extra database messages in the job log? This can help determine what 
database tables may need additional indexes created, for example. Have 
you looked into any of those tools and techniques?
What version of OS/400 or IBM i are you running? Are your database 
tables and views mainly defined with SQL DDL, or are they physical files 
and logical files defined with DDS? There are many IBM Redbooks on 
tuning your database and applications, as well as other vendor tools.
Search the MIDRANGE-L archives; you should find many discussions of 
tuning database queries to improve performance. Or search with google 
for: "tuning i5/os DB2 sql" (without quotes)...
HTH
Mark S. Waterbury
> On 1/16/2011 4:30 PM, Tomasz Skorża wrote:
Thabks for answer about memories ... I have one more question.
Is it possible to estimate (even approximately) how much the system efficiency will grow together
with memory grow?
4 GB to 8GB? 10%? 20% more efficiency?
4 to 16?
4 to 32?
I'm interested in time of get result of SQL queries.
Regards
Tomek
As an Amazon Associate we earn from qualifying purchases.