Sybase 12.4.2 Example of setting buffer cache sizes, 426, Other considerations, Memory available

Overview of memory use

Other considerations

Adaptive Server IQ buffer cache sizes may differ from one database to the next based on how you use it. For maximum performance, you need to change the settings between inserting, querying the database, and mixed use. In a mixed- use environment, however, it is not always feasible to require all users to exit the database so that you can reset buffer cache options. In those cases, you may need to favor either load or query performance. When possible, define the cache sizes before doing any work in the database.

The buffer cache and memory overhead guidelines also may differ between platforms. See your Adaptive Server IQ Installation and Configuration Guide for any other issues.

Example of setting buffer cache sizes

The following table provides a list of the factors that may consume memory for your system and how much remains for your main and temp buffer caches. It assumes that the system has 1GB of physical memory, no other significant applications on the hardware other than running Adaptive Server IQ, and only one active database at a time. The table makes a distinction in the storage type (raw versus “cooked”) and the way the database is accessed (queries or inserts).

Table 12-1: Memory available for buffer caches (Example)

				Memory available
		Memory available		using "cooked" file
		using raw partitions		systems
	Amount
Memory Use	Used	Queries	Inserts	Queries	Inserts
Total amount of physical memory		1000	1000	1000	1000
available (approximate in MB)
Operating system use assuming a	60a	940	940	940	940
minimum amount for a UNIX
system
Overhead from using "cooked" file	278			647	647
system: 30% of remaining memory
from step 3 above
Overhead for number of active	100	825		547
users: approximately 30 connected
users but only about 10 active at
10MB each
Overhead for inserts from flat files	97		828		550
assuming a 200-byte record size and
default settings

426