Support Database Parameters

Use

Most support database parameters are for troubleshooting. In normal operation, it is not necessary to change the support database parameters. Changing support database parameters requires very detailed knowledge of the database system. For information about other database parameters, see General Database Parameters, Special Database Parameters and liveCache Database Parameters.

Changes to database parameters do not take effect until the database instance is restarted.

You can also change some database parameters while the database is running. In addition you can choose how long the change is to be effective for these database parameters:

¦      Until the next restart

¦      Not until after the next restart

¦      Immediately and permanently

Support Database Parameters

Parameter

Meaning

How can I change this Parameter?

_BACKUP_HISTFILE

Name of the history file for data and log backups

It cannot be altered.

_BACKUP_MED_DEF

Name of the file containing the definitions of the backup templates

It cannot be altered.

_EVENTFILE

Name of the file in which the database system logs internal events

Changes only apply after restart

_EVENSIZE

Size of the file in which the database system logs internal events

Changes only apply after restart

_IDXFILE_LIST_SIZE

Number of temporary intermediate result files in the case of parallel indexing

The database system indexes large tables using multiple server tasks. These server tasks write their results in temporary files. When the number of these files reaches the value of this parameter, the database system has to merge the files before it can generate the actual index. This results in a decline in performance.

Can be changed during operation

_IOPROCS_PER_DEV

Number of threads that the database system can use for asynchronous I/O operations (per data volume and system)

Changes only apply after restart

_KERNELDIAGFILE

Name of the log file of the kernel

Changes only apply after restart

_KERNELDUMPFILE

Name of the dump file that is written by the kernel when the system crashes

To this file, the database system writes, among other things, the contents of the data cache and the converter. Size your system so that sufficient memory space is available for this file (approximately CACHE_SIZE + 10%). Note also parameter DIAG_HIST_NUM.

Changes only apply after restart

_KERNELTRACEFILE

File in which the kernel writes the trace entries

The kernel only writes trace entries if you previously activated the database trace.

Changes only apply after restart

_MAXEVENTTASKS

Maximum number of event tasks

The database system uses event tasks both for Event Dispatcher and for the DBM commands event_wait, event_receive, event_available, auto_extend and auto_update_statistics. These commands can be active simultaneously for this database instance.

The default value of the Database Manager for this parameter is 2.

Changes only apply after restart

_MAXTASK_STACK

Size of the stack that is used by the user tasks

Changes only apply after restart

_MINREPLY_SIZE

Minimum size of the memory that is available in a package (shared memory segment) for the reply

0: For the application as well as for the kernel, the whole package is available for the request/reply.

Changes only apply after restart

_MULT_IO_BLOCK_CNT

see DATA_IO_BLOCK_COUNT and LOG_IO_BLOCK_COUNT

_PACKET_SIZE

Size of the packages (shared memory segments) into which SQL statements and data are transmitted

A package consists of a request part and a reply part.

Changes only apply after restart

_READAHEAD_BLOBS

Number of pages from which large LONG values are imported in advance by additional server tasks

If a LONG value is too large to be transferred to the client in a single request package, the database system splits it into several request packages. To increase performance, server tasks can import further parts of the LONG value while the first request package is being sent.

Can be changed during operation

_RESTART_TIME

Minimum time between two savepoints (in seconds); corresponds to the time that is required for a restart after a system crash

Beyond this, the database system always writes a savepoint in the following cases:

?     When you have created an index

?     When the number of pages released by a savepoint is larger than the number of free blocks in the data volumes

Changes only apply after restart

_RTEDUMPFILE

File in which the kernel writes information about the runtime environment if the system crashes

Changes only apply after restart

_SERVERDB_FOR_SAP

Specifies whether the database instance is being used in an SAP system

Can be changed during operation

_TASKCLUSTER_01 to _03

These parameters describe how the database system distributes user tasks to the threads. Only change these parameters after consulting support.

Changes only apply after restart

_USE_ASYNC_IO

Specifies whether operating system functions or special I/O threads are used for asynchronous I/O operations

Changes only apply after restart

_USE_IOPROCS_ONLY

YES: I/O operations are executed only by special I/O threads

NO: the kernel decides whether a task is to execute an I/O operation itself or give the I/O operation to a special I/O thread

Can be changed during operation

ALLOW_MULTIPLE_SERVERTASK_UKTS

Specifies whether the database system distributes the server tasks to the available user kernel threads or if they all run in the same user kernel thread

Changes only apply after restart

COLUMNCOMPRESSION

Determine whether the column values can be stored in variable length or not.

Default value: YES (columns have variable length, so the values can be compressed if necessary)

Only for columns that are not primary key columns

Can be changed during operation

DATA_IO_BLOCK_COUNT

Block size that the database system uses when writing data pages from the data cache to the data area.

The optimum block size depends on the hardware and software that is used.

Changes only apply after restart

ENABLE_SYSTEM_TRIGGERS

Defines whether system triggers are called when the database is restarted

Changes only apply after restart

EXPAND_COM_TRACE

Specifies whether memory space is reserved when COM trace files are created

Changes only apply after restart

INIT_ALLOCATORSIZE

Initial size of the working memory that is reserved at the start of a database session

Changes only apply after restart

LOAD_BALANCING_DIF

Only for load balancing; specifies by how much longer the task to be moved had to wait in its user kernel thread than the task that has been waiting the longest in the target user kernel thread (as a percentage)

Changes only apply after restart

LOAD_BALANCING_EQ

For load balancing; specifies which time delay should be regarded as equal when comparing waiting tasks (as a percentage)

Changes only apply after restart

LOG_IO_BLOCK_COUNT

Block size that the database system uses when writing log pages from the log queues to the log area

The optimum block size depends on the hardware and software that is used.

Changes only apply after restart

LOG_QUEUE_COUNT

Number of log queues

=0: Number calculated from the MAXCPU parameter

>0: Value of the USED_LOG-QUEUE_COUNT parameter is adopted

Changes only apply after restart

MAX_SERVERTASK_STACK

Maximum size of the stack that is used by the server tasks

Changes only apply after restart

MAX_SPECIALTASK_STACK

Maximum size of the stack that is used by special tasks (all except user tasks and server tasks)

Changes only apply after restart

MAXPAGER

Maximum number of pagers; calculated by the database system from MAXDATAVOLUMES, among other parameters

Not changeable

You can override this parameter with XP_MAXPAGER.

MAXVOLUMES

Maximum number of data and log volumes including mirrored volumes; calculated by the system

Changes only apply after restart

MINI_DUMP

Defines whether the system writes the memory dumps knlmini.dmp and srvmini.dmp (for postmortem debugging) and how much information these dumps contain

DISABLED: the system does not write any memory dumps

NORMAL (default value): the system writes all stacks and system handle information in the memory dump

FULL: the system writes all stacks and data segments in the memory dump

Changes only apply after restart

OMS_STREAM_TIMEOUT

Maximum wait time for all database sessions that may elapse until the reply to an OMS stream request to a client (in seconds)

Can be changed during operation

OPTIM_CACHE

Defines whether the database system determines the search strategy only once or every time it executes a parsed SQL statement

For prepared statements with parameters, it may be sufficient to determine the search strategy only once.

Can be changed during operation

OPTIMIZE_OPERATOR_JOIN

Defines whether the database system uses optimized join implementation (saves resources)

Can be changed during operation

SET_VOLUME_LOCK

Specifies whether the database system sets a lock when a volume is added, which prevents the same volume from being added again later

In the following cases it may make sense to set this parameter to NO:

?     NFS-mounted volume

?     Hot Standby System

Changes only apply after restart

SHOW_MAX_STACK_USE

Logs the maximum stack usage of each task in the kernel log

Only set this parameter to YES for diagnosis purposes as it compromises performance.

Changes only apply after restart

SUPPRESS_CORE

Defines whether the database system suppresses core dumps of the kernel

Changes only apply after restart

SYMBOL_DEMANGLING

Defines whether the database system performs C/C++ demangling

Changes only apply after restart

UPDATESTAT_PARALLEL_SERVERS

Defines how many parallel server tasks the database system uses for updating the SQL Optimizer statistics

Value -1: update of statistics is done sequentially
Value 0: database system decides how many parallel server tasks should be used

Value n>0: database system uses a maximum of n parallel server tasks

Can be changed during operation

USE_OPEN_DIRECT

YES: the database system uses the O_DIRECT flag when opening volumes (if this flag is supported by the file system). You use this flat to configure that the operating system does not use a separate cache for I/O operations.

Note that this flag is ignored for Linux kernel < 2.4.18.

Changes only apply after restart

USE_SYSTEM_PAGE_CACHE

Defines whether the database instance uses the system page cache for buffering memory pages that are no longer required

Changes only apply after restart