Sybase 12.4.2 User Manual

Overview of memory use

 

Other considerations, such as hardware and network analysis, can locate 
bottlenecks in your installation.

Adaptive Server IQ uses memory for several purposes:

•

Buffers for data read from disk to resolve queries

•

Buffers for data read from disk when loading from flat files

•

Overhead for managing connections, transactions, buffers, and database 
objects

The sections that follow explain how the operating system supports IQ's use of 
memory, how IQ allocates memory for various purposes, how you can adjust 
the memory allocations for better performance, and what you may need to do 
to configure the operating system so that enough memory is available for IQ.

When there is not enough memory on your system, performance can degrade 
severely. If this is the case, you need to find a way to make more memory 
available. Like any RDBMS software, Adaptive Server IQ requires a lot of 
memory. The more memory you can allocate to Adaptive Server IQ, the better. 

However, there is always a fixed limit to the amount of memory in a system, 
so sometimes operating systems can have only part of the data in memory and 
the rest on disk. When the operating system must go out to disk and retrieve 
any data before a memory request can be satisfied, it is called paging or 
swapping. The primary objective of good memory management is to avoid or 
minimize paging or swapping.

The most frequently used operating system files are swap files. When memory 
is exhausted, the operating system swaps pages of memory to disk to make 
room for new data. When the pages that were swapped are called again, other 
pages are swapped, and the required memory pages are brought back. This is 
very time-consuming for users with high disk usage rates. In general, try to 
organize memory to avoid swapping and, thus, to minimize use of operating 
system files.