General Extended Security Support By Name By Class History

SAP MaxDB Database Parameters (7.9.10.09, Extended)


AutoTruncateBackupHistory

default: 0

change: DIRECT

new in 7.9.10.05

 This parameter is a DBM Server configuration parameter.

 If "AutoTruncateBackupHistory" is set to a value of 

	182 <= "AutoTruncateBackupHistory" < 3650

 the backup history files are truncated at each complete data backup so that
 at least a period of the given value (days) remains in the backup history
 files.
 The truncation takes place as defined by the DBM Server command 
 "db_truncatehistory DATE ( - 'AutoTruncatebackupHistory')".

 A value of '0' means no automatic truncation is active.

 The default value is 0 days, i.e. the feature is deactivated.
  

CatalogCacheSupply

change: RUNNING

new in 7.9.09.01

 CatalogCacheSupply is the reserved space for all task-specific catalog caches.
 For each user task a minimum part of the catalog cache is reserved.
 (TaskSpecificCatalogCacheMinSize) Each task may extend its catalog cache unless all
 task-specific catalog caches together do not exceed the catalog cache supply.
 The catalog cache supply influences the performance. The best
 performance can be achieved at a catalog cache hit rate of 100%.
 Use the database monitoring to get information about the catalog
 cache hit rate.

 The lower and upper limits are:
  (MaxUserTasks + PreservedDBAnalyzerTasks + MaxServerTasks + MaxUtilityTasks) * 2 * TaskSpecificCatalogCacheMinSize <= CatalogCacheSupply
          and 30000 <= CatalogCacheSupply <= 2147483647

 Default calculation, can be overridden
  CatalogCacheSupply := (MaxUserTasks + PreservedDBAnalyzerTasks + MaxServerTasks + MaxUtilityTasks) * 2 * TaskSpecificCatalogCacheMinSize but at least 30000

           TaskSpecificCatalogCacheMinSize depends on CommandBufferSize

 (4 bytes integer)


CheckAllocator

default: depends on operating system or instance type

change: RUNNING

  defines the degree of heap checking
  0 disables heap checking
  
  The value is a bit field and can be set as follows:
  
    CheckAllocator & 1   - places a pattern after each used chunk. It will be checked when the chunk is freed
    CheckAllocator & 2   - fill the chunks with a pattern when it is freed
    CheckAllocator & 4   - double bookkeeping of used chunks
    CheckAllocator & 8   - check binary tree of big free chunks
    CheckAllocator & 16  - check exclusive usage of unsynchronized heaps
    CheckAllocator & 32  - delay the reuse of freed chunks (delayed free)
    CheckAllocator & 64  - check in use chunks (kernel only)
    CheckAllocator & 128 - check in use OMS chunks

 (4 bytes integer)

ClusterCompressionFillThreshold

default: 10

change: RUNNING

  The lower and upper limits are:
  0% <= ClusterCompressionFillThreshold <= 50%
  
  Clusters with up to ClusterCompressionFillThreshold occupied blocks will be considered for compression. Percent value of the DataIOClusterSize
  
  0 will disable compression

ClusterWriteThreshold

default: 80

change: RUNNING

  The lower and upper limits are 0% and 100%
  
  Minimum number of blocks to write into a cluster. Percent value of the DataIOClusterSize

Comment

change: OFFLINE

 Optional comment for the database

 (64 byte character)

ConverterStripes

default: 8

change: OFFLINE

 Number of autonomous units of the component converter in
 which parallel working is done.

 Note that the number of ConverterStripes are determined
 by a calculation regulation which depends on the capacity
 of the DATA_AREA (parameter DataVolumeSize*). 

 The lower and upper limits are:
     1 <= ConverterStripes <= 64

            DATA_AREA_SIZE <=   62 GB  --> ConverterStripes =  8
    62 GB < DATA_AREA_SIZE <=  125 GB  --> ConverterStripes = 16
   125 GB < DATA_AREA_SIZE <=  250 GB  --> ConverterStripes = 32
   250 GB < DATA_AREA_SIZE <=  500 GB  --> ConverterStripes = 48
   500 GB < DATA_AREA_SIZE             --> ConverterStripes = 64

 (2 bytes integer)

ConverterVolumeIdLayout

default: 8

change: OFFLINE

 Number of bits in a four byte converter block address
 reserved for the logical devive number which identifies
 the data volumes. The rest of the four byte block
 address is used for the device offset.
 The greater the ConverterVolumeIdLayout the more data volume
 could be managed.
 Note that this reduces the capacity of usable data volumes!
 This value is not changeable within the lifetime of the data
 base without making a recovery.

 Example: If ConverterVolumeIdLayout is set to the default value of 8 
		  then the data base is able to manage as far as 256 data volumes, 
		  whereby the maximum capacity of one data volume is 128 GB.

          If the ConverterVolumeIdLayout is set to 6 the data base is able
          to manage up to 64 data volumes, whereby the capacity of
          one data volume is 512 GB.

 The lower and upper limits are:
     6 <= ConverterVolumeIdLayout <= 12


DataCacheStripes

change: OFFLINE

 Number of autonomous units of the data cache in which
 parallel working is done.

 Note that the number of DataCacheStripes are determined
 by a calculation regulation which depends on the capacity
 of the parameter CacheMemorySize. 

 The lower and upper limits are:
     1 <= DataCacheStripes <= 1024

            CacheMemorySize <=    10000  --> DataCacheStripes =    8
    10000 < CacheMemorySize <=    50000  --> DataCacheStripes =   16
    50000 < CacheMemorySize <=   100000  --> DataCacheStripes =   32
   100000 < CacheMemorySize <=   150000  --> DataCacheStripes =   48
   150000 < CacheMemorySize <=  1000000  --> DataCacheStripes =   64
  1000000 < CacheMemorySize <=  4000000  --> DataCacheStripes =  128
  4000000 < CacheMemorySize <=  8000000  --> DataCacheStripes =  256
  8000000 < CacheMemorySize <= 12000000  --> DataCacheStripes =  512
 12000000 < CacheMemorySize              --> DataCacheStripes = 1024

 (2 bytes integer)


DateTimeFormat

default: "INTERNAL"

possible values: "INTERNAL" "ISO" "EUR" "JIS" "USA"

change: RUNNING

 Alterable default format of date, time and timestamp values for non-JDBC
 and non-ODBC applications

			date format	time format timestamp format	
			__________________________________________________
 'INTERNAL'	YYYYMMDD	HHHHMMSS	YYYYMMDDHHMMSSMMMMMM
 'ISO'		YYYY-MM-DD	HH:MM:SS	YYYY-MM-DD HH:MM:SS.MMMMMM
 'EUR'		DD.MM.YYYY	HH.MM.SS	YYYY-MM-DD-HH.MM.SS.MMMMMM
 'JIS'		YYYY-MM-DD	HH:MM:SS	YYYY-MM-DD-HH.MM.SS.MMMMMM
 'USA'		MM/DD/YYYY	HH:MM AM	YYYY-MM-DD-HH.MM.SS.MMMMMM



DeadlockDetectionLevel

default: 4

change: RUNNING

 Maximum search level for deadlock detection.

 The lower and upper limits are:
 0 <= DeadlockDetectionLevel <= 10000

 0   : deadlock detection is disabled, deadlocks are only
       cancelled by the RequestTimeout.

 > 0 : deadlocks not detected by the deadlock detection within the given
       search level are only cancelled by the RequestTimeout.

 (4 bytes integer)

DefaultCodePage

default: "ASCII"

possible values: "ASCII" "UNICODE"

change: RUNNING

 Default code taken as code attribute for CHAR columns
 defined without code attribute, for open systems usually ASCII code

 DEFAULT CODE = 'ASCII' or 'UNICODE'

 (UNICODE is only valid for a unicode database)

 (char(18))

DeterministicFunctionExecution

default: "EXECUTE"

possible values: "EXECUTE" "PREPARE"

change: RUNNING

 DeterministicFunctionExecution 'EXECUTE' or 'PREPARE'

 'EXECUTE': Deterministic result of user defined functions will be calculated at execution time

 'PREPARE' : Deterministic result of user defined functions will be calculated at prepare time

DiagnoseHistoryCount

default: 2

change: OFFLINE

 Number of diagnostic file histories written after an abnormal database shutdown

 Number of histories in the DiagnoseHistoryPath.
 At database start some important diagnostic files
 are saved to a history directory named 
 '_' which is created under the 
 DiagnoseHistoryPath located in the RunDirectoryPath.
 
 If the number of histories is reached the oldest
 history is deleted and the new one is written instead.

 The value has to be greater than 0 to insure that
 at least one backup of diagnostic files can be initiated
 after an abnormal database shutdown.

 The default number of histories is 2.

 (4 bytes integer)

DiagnoseHistoryPath

change: OFFLINE

 This directory holds different kinds of diagnostic
 histories which are located under further subdirectories.
 
 First of all the histories of diagnostic files after
 a abnormal database shutdown are located in this directory.
 
 Furthermore a history of the last kernel message files
 is also located in this directory.
 
 The default directory for this is "DIAGHISTORY"
 which is created under the specified RunDirectoryPath.

 (256 bytes character)

EnableIndexOnlyStrategy

default: "YES"

possible values: "YES" "NO"

change: RUNNING

 EnableIndexOnlyStrategy 'YES' or 'NO'

 'YES': avoid base table access with a primary key fetched from an index
        ( if possible )
 'NO' : access base table with primary key fetched from an index

 (char(8))

EnableLogBackupToPipe

default: "NO"

possible values: "YES" "NO"

change: OFFLINE

EnableLogBackupToPipe = 'YES' or 'NO'

Allow the usage of pipes for log backup similar to files. 
This includes the automatic release of log after the backup 
although the database can not check if the backup was successful.

For the usage of pipes with 3rd party backup tools set this 
parameter to NO. With 3rd party backup tools the log will be
released automatically after the database checked the success 
of the backup. 

Possible values are:

  'YES' : log will be freed automatically without check of the backup into the pipe
  'NO'  : log will be not freed automatically (except for 3rd party backup tools)

(8 byte character)

EnableQueryRewrite

default: "YES"

possible values: "NO" "YES"

change: RUNNING

 EnableQueryRewrite 'NO' or 'Yes'

 'NO'  : No query rewrite will be done.
 'YES' : Query will be rewritten if possible.


ExclusiveLockEnqueueThreshold

default: 2

change: RUNNING

 The tasks will skip the active waiting phase during 
 the exclusive-lock collision if the waiting queue
 filling exceeds this threshold.

 The limits are:
     2 <= ExclusiveLockEnqueueThreshold

 (4 bytes integer)

ExclusiveLockFavorPrioTasks

default: "DEFAULT"

possible values: "YES" "NO" "DEFAULT"

change: RUNNING

 This parameter specifies whether or not a priority task that does
 not obtain a critical region at once is to be favored when another
 task frees this region.

 ExclusiveLockFavorPrioTasks = 'YES', 'NO' or 'DEFAULT'

 'YES' means: If a priority task tries to obtain a region used by
              another task (collision), the priority task will add
              itself to a special queue in order to obtain the region
              as soon as the other task frees the region.

 'NO'  means: see parameter MaxExclusiveLockCollisionLoops
 'DEFAULT'  means:
 IF MaxCPUs > 1
 THEN
       act as if set to 'YES'
 ELSE
       act as if set to 'NO'

 (char(8))

ExclusiveLockRescheduleThreshold

default: -1

change: RUNNING

 Maximum number of attempts a task can make to obtain a region
 without being interrupted by another task in this UKT.
 When this number is reached without success, the task
 interrupts itself and falls asleep.

 If this parameter is set to -1, a default value is chosen by
 the kernel. The default value is 300 for MaxCPUs = 1 and 3000
 for MaxCPUs > 1.

 If LockRescheduleInterval has been set (value not equal to 0),
 the lower and upper limits are:
     0 <= ExclusiveLockRescheduleThreshold <= 10000

 Otherwise:
     100 <= ExclusiveLockRescheduleThreshold <= 10000

 (4 bytes integer)

FileDirectorySpinlockPoolSize

default: 10

change: OFFLINE

 This parameter controls the number of spinlocks available for the
 file directory spinlock pool.

 The file directory uses a large number of reader/writer locks to
 provide maximum concurrency during operation. When changing states of
 any of those reader/writer locks a spinlock has to be used to protect
 that lock's internal state change. Since spinlocks are expensive in
 terms of memory and cache usage every reader/writer lock is mapped to
 one specific spinlock from a spinlock pool. This means that several
 reader/writer locks share the same spinlock to protect their internal
 state.

 If you have many tables/objects and a high degree of concurrency in
 your database the default setting of FileDirectorySpinlockPoolSize might
 lead to a bottleneck because reader/writer lock accesses have to
 serialize on the available spinlocks. You can determine if this is the
 case by checking the values for column COLLISIONCOUNT in rows with
 name 'FileDir%' in the system view SPINLOCKSTATISTICS. If these seem
 unusually high for most of the available spinlocks you should
 increase FileDirectorySpinlockPoolSize.

 Configuring too high a FileDirectorySpinlockPoolSize will not improve the
 database performance but only waste precious resources.

FormatDataVolume

default: "YES"

possible values: "YES" "NO"

change: OFFLINE

  FormatDataVolume = 'YES' or 'NO'

In order to speed up the installation process
on personal development systems the formatting
of data volumes (type file) can be suppressed.

It is not recommended to suppress the formatting
of data volumes on productive systems. The formatting process
occupies the desired space on disk. Therefor omitting it
may lead to a system crash during runtime if the system
is low on disk space.

Possible values are:
 'YES' : Formatting of data volumes is performed (default)
 'NO'  : Formatting of data volumes is suppressed


HashedResultsetCacheSize

default: 65536

change: RUNNING

 The maximum cache size in KB used for each hashed resultset

 The lower and upper limits are:
      32768 (=32MByte) <= HashedResultsetCacheSize <= 8388608 (=8GByte)


HashJoinSingleTableMemorySize

default: 4000

change: RUNNING

 The maximum table size in KB for which hash joins will be executed.

 The lower and upper limits are:
     0 <= HashJoinSingleTableMemorySize <= HashJoinTotalMemorySize

 If HashJoinSingleTableMemorySize = 0 then no hash tables will be created during
 join execution.

HashJoinTotalMemorySize

default: 24000

change: RUNNING

 The maximum memory in KB that can be used for hash joins at any time.
 As there can be multiple hash joins running at the same time, the amount
 of memory used for all hashes might become excessive if it is unlimited.
 This parameter sets the upper limit for the memory provided for all
 hash joins that are running in parallel. If during join execution a join
 transition qualifies for a hash join but the overall memory used for all
 hash joins would be more than HashJoinTotalMemorySize a regular join will
 be executed instead.

 If HashJoinTotalMemorySize = 0 then no hash joins will be executed.

HighResolutionTimer

default: depends on operating system or instance type

possible values: "CPU" "GETTIMEOFDAY" "PERFCNTR" "TICKS" (values depend on operating system)

change: OFFLINE


Specifies the high resolution timer type used by the database kernel.

Possible values are:

 CPU:
   Direct access to the CPUs hight resolution counters.

 GETTTIMEOFDAY:
	Using system gettimeofday() system call.
    
 PERFCNTR:
	Using 'QueryPerformanceCounter' functionality.

 TICKS:
	Using system clock ticks.
	
In some very rare situations the result that is returned by the CPU counters may 
unexpectedly leap forward from time to time. This leap may represent several seconds
due to some hardware design defects or different CPU speeds in multi CPU environments.
This parameter allows to overrule the default usage of CPU timers which is faster.
	
(char(16))

     

HotStandbyDatabaseName?

change: OFFLINE

Specifies the database name in a high availability system using no shared resources
used to connect between master and standby server.
A three digit number is used to specify the names leading to a range from 
HotStandbyDatabaseName001 up to HotStandbyDatabaseName999.
The default for additional databases in a high availability system will be the name
of the master database. 
This parameter does not apply to high availability systems using a shared log area!
 (char(64))

HotStandbyDelayTime?

change: OFFLINE

In a high availability configuration using a shared log area, this parameter specifies 
the local node delay time in seconds.
A three digit number is used to specify the specific nodes leading to a
range from HotStandbyDelayTime001 up to HotStandbyDelayTime999
This parameter does not apply to high availability systems using no shared resources!
 (char(64))

HotStandbyNodeName?

change: OFFLINE

Specifies the local node name of a high availability node used to connect between master
and standby server. A three digit number is used to specify the nodes leading to a
range from HotStandbyNodeName001 up to HotStandbyNodeName999. HotStandbyNodeName001 has the special role of being
default master node.

 (char(64))

HotStandbyStorageDLLPath

change: OFFLINE

 Only needed for a high availability configuration, that means if at least HotStandbyNodeName001 is configured too.

 When high availability using a shared log area is configured, the storage access is encapsulated in a 
 separated dynamic link library (DLL).
 This parameter allows to setup the name of that DLL. Different DLLs for different storage
 system are possible. The DLL is itself loaded dynamically on demand in high availability code sections.

 Example HotStandbyStorageDLLPath names are "libhsscopy" and "libhssemc".

 For a high availability system without shared log area, no DLL is used. To indicate such a system, leave
 the value of this parameter blank (default).

 (256 bytes character)

HotStandbySyncInterval

default: 50

change: OFFLINE

 Specifies the time interval in seconds between two synchronize commands of the master node in a high availability configuration.

 In a high availability configuration using a shared log area the master node sends synchronize commands 
 to the standby nodes.
 The minimum time interval between two successive synchronize commands is defined by the  parameter
 HotStandbySyncInterval.  The synchronize command is initiated by the timeout task. This leads to an
 inaccuracy  of this parameter because it is only tested once during the runtime of the timeout task.

 The lower and upper limits are:
     0 <= HotStandbySyncInterval <= 36000

 This parameter does not apply to high availability systems using no shared resources!
 (4 bytes integer)

IndexlistsMergeThreshold

default: 500

change: RUNNING

 Strategies for merging index lists produce a temporary index which
 implies space and time consumption.
 To minimize these disadvantages those strategies are only used
 when the number of index pages involved is less than or equal
 to IndexlistsMergeThreshold. As an alternative the strategy which only works
 on the base table is used.

 The lower and upper limits are:
     0 <= IndexlistsMergeThreshold

 (4 bytes integer)

JoinSearchLevel

default: 0

change: RUNNING

 The lower and upper limits are:
 0 <= JoinSearchLevel <= 9

 A search level determines time and effort of join sequence search.
 With level 9 most of all possible join sequences will be evaluated.
 A simple algorithm will be used with level 1. Level 0 means graded search
 level according to participated tables. Up to JoinSearchTableThreshold9 tables
 level 9 will be used, up to JoinSearchTableThreshold4 tables level 4 will
 be used.

 (4 bytes integer)

JoinSearchTableThreshold4

default: 16

change: RUNNING

 The lower and upper limits are:
 2 <= JoinSearchTableThreshold4 <= 64

 Level 4 sequence search for joins with up to JoinSearchTableThreshold4 tables.
 This threshold will be used with automatic join search level
 ( JoinSearchLevel = 0 )

 (4 bytes integer)

JoinSearchTableThreshold9

default: 8

change: RUNNING

 The lower and upper limits are:
 2 <= JoinSearchTableThreshold9 <= 64

 Level 9 sequence search for joins with up to JoinSearchTableThreshold9 tables.
 This threshold will be used with automatic join search level
 ( JoinSearchLevel = 0 )

 (4 bytes integer)

JoinTableBufferSize

default: 256

change: RUNNING

 Specifies the minimal amount of buffer size (in KB) per user task for 
 table buffers used during join execution

 (4 bytes integer)

KernelMessageFileSize

default: 10000

change: OFFLINE

 Size of the file which contains all messages, warnings, and errors
 of the database kernel.

 The file is written cyclically.

 The lower limit is:
     800 <= KernelMessageFileSize

 (2 bytes integer)

KernelMessageHistoryCount

default: 20

change: OFFLINE

 Number of kernel message file backups.
 
 At database start the last kernel message file 'KnlMsg'
 is moved to a history directory named 
 'History_KnlMsg' which is created under the 
 DiagnoseHistoryPath located in the RunDirectoryPath.
 
 A timestamp is added to the name of the kernel message file 
 when it is moved to the history.
 
 If the number of message files in the history is reached the oldest
 message file is deleted and the new one is written instead.

 The value has to be greater than 0 to insure that
 at least one backup of the kernel message file can be 
 written.

 The default number of files in the history is 20.

 (4 bytes integer)

LoadBalancingCheckInterval

default: 1

change: RUNNING

Load balancing means distributing tasks across multiple
threads (UKTs) within the database kernel.
The parameter LoadBalancingCheckInterval determines the
the time how often a check will be made
to find a task which have to move to another thread (UKT).

Possible values are:
 0:
   No load balancing will be made.
 between 1 and 600: 
   Time in seconds, how often a balancing check is made.

(integer)

LocalRedoLogBufferSize

default: 0

change: OFFLINE

 If this is set to a value greater than 0, a local buffer for each transaction is allocated.
 The given value is the buffer size in Bytes. The buffer is used to store redo entries
 before flushing them at once to the log volume.
 If the number of log queues (see parameter LogQueues) is greater than 1, no local
 redo buffer can be used.

LogQueues

default: 0

change: RUNNING

 The parameter LogQueues defines the number of log queues, which are used for writing onto the log volume.
 If set to 0 the value will be set to the value of parameter MaxLogQueues.
 The lower and upper limits are:
    0 <= LogQueues <= MaxLogQueues
 (2 bytes integer)

 More than 1 LogQueue is only possible if parameter LocalRedoLogBufferSize is set to 0.
 Otherwise LogQueue is set to 1.

LogQueueSize

default: depends on operating system or instance type

change: OFFLINE

 The LogQueueSize defines the size of a single log queue (see parameter LogQueues for 
 the number of log queues) in pages.

 By adjusting the LogQueueSize peak loads can be avoided and a larger LogIOClusterSize can be used
 when writing to the log.

 Use the database monitoring to get information about the number of
 LOG QUEUE OVERFLOWS and LOG QUEUE MAX USED ENTRIES.

 The lower and upper limits are:
     LogQueueSize < LogVolumeSize*
     AND
     LogQueueSize * MaxLogQueues < LogVolumeSize*
     AND
     LogQueueSize >= 2 * LogIOClusterSize

 (2 bytes integer)

MaxEvents

default: 150

change: OFFLINE

 Maximum number of events cached by the kernel for being processed by the
 Database Manager.

 The lower limit of 0 disables the caching of any event. It is not recommended
 to disable the caching of events, if not for minimizing memory consumption since
 the database manager can only process cached events. If no events are cache the
 database manager cannot report any events.

 (2 bytes integer)

MaxExclusiveLockCollisionLoops

default: -1

change: RUNNING

 Maximum number of attempts made to obtain a region used by another
 task (collision).
 If the attempt fails, the UKT falls asleep (semop).

Set this parameter > 0 ONLY in combination with UseableCPUs > 1.

 If this parameter is set to -1, a value is chosen by the kernel
 as follows:
    IF MaxCPUs = 1
    THEN
       MaxExclusiveLockCollisionLoops := 0
    ELSIF MaxCPUs < 8
       MaxExclusiveLockCollisionLoops := 1000
    ELSE
       MaxExclusiveLockCollisionLoops := 100000

 (4 bytes integer)

MaxKernelMessageArchiveSize

default: 10000

change: RUNNING

 If this parameter is set to a value other than 0, it defines a maximum size (in KB) for 
 the file KnlMsgArchive (containing archived messages of the database kernel).
 If this size (in KB) is reached, the file is closed and renamed to KnlMsgArchive_.
 A new file KnlMsgArchive is then created for any coming error messages generated.
 (2 bytes integer)

MaxLogWriterTasks

default: 0

change: OFFLINE

Used for defining the number of log writer tasks for multiple log partitions

MaxParallelLiveCacheTraceFiles

default: depends on operating system or instance type

change: OFFLINE

 Maximum number of trace files which can be opened in parallel (liveCache Instances only)

 The limits are:
      64 <= MaxParallelLiveCacheTraceFiles <= 120

 (2 bytes integer)

MaxParameterStringLiteralSize

default: 200

change: RUNNING

	The maximum length of a string passed as unspecific parameter argument to functions.
	These parameter arguments are used mostly in LPAD, RPAD, COALESCE system functions.

MaxRetentionTime

default: 480

change: RUNNING

 Forces the garbage collector to remove consistent views which are
 older than MaxRetentionTime.
 Time is specified in minutes.

 The lower and upper limits are:
     MinRetentionTime <= MaxRetentionTime

 If MaxRetentionTime = 0 then no consistent view is cancelled by the
 garbage collector because of being too old.

 If MinRetentionTime = 0 AND MaxRetentionTime = 0 no consistent views
 will ever be cancelled by garbage collection. (4 bytes integer)


MaxServerTasks

default: 0

change: OFFLINE

 Server tasks are used for restart, backup/restore, savepoint, create index and update statistics operations and savepoints.

 Overconfiguration exceeding the actual requirement results in an excessive
 demand for address space.

  MaxServerTasks >=  ReservedServerTasks + 
                     MinCheckDataParallelServerTasks + 
                     1 ( = used for ASYNC DROP )

 (2 bytes integer)

MaxUtilityTasks

default: 1

change: OFFLINE

new in 7.9.10.00

 Maximum number of simultaneously active utility tasks.
 
 The utility task(s) can be used in cases where all user tasks are blocked 
 to solve database- or log-full situations.   

 (2 bytes integer)

MicroSecondTimer

default: "HighResolutionTimer"

possible values: "HighResolutionTimer" "SystemTimer"

change: OFFLINE


Specifies the timer type used for time stamps etc.

Possible values are:

 HighResolutionTimer:
   Using the fast high-resolution-timer  

 SystemTimer:
    Using the standard operating system timer.
    
(char(40))

MinRetentionTime

default: 60

change: RUNNING

 Forbids the garbage collector to remove consistent views which are
 younger than MinRetentionTime.
 Time is specified in minutes.

 The lower and upper limits are:
     0 or 3 < MinRetentionTime <= 1440 (one day)

 If MinRetentionTime = 0 then no consistent view is cancelled by the
 garbage collector because of MinRetentionTime even with resource
 shortage.

 If MinRetentionTime = 0 AND MaxRetentionTime = 0 no consistent views
 will ever be cancelled by garbage collection.

 (4 bytes integer)

OfficialNodeName

change: RUNNING

 OfficialNodeName

 The official node name of the host of the database
 kernel may be ambiguous or even misleading.
 Especially in a cluster configuration with several network
 interfaces the decision which name is to be used by clients
 to establish a remote connection is problematic. Each
 interfaces must defines its own 'hostname'.

 If this parameter is set, the database kernel always
 uses it instead of the default value.

 If it is not set (which is default) the database
 kernel uses the nodename retrieved by

	'uname()' call (UNIX)

        Since some administrators misuse 'uname' for a nickname
        of their computer, it is a way to keep the nickname
        and still allow clients to connect...

 or

	official hostname (NT).

        The official hostname is retrieved by first
        using GetComputerName()
        followed by get_host_by_name()
        and finally get_host_by_addr().
        This results normally in the first name given
        in the alias list in /etc/hosts where the
        computer name is found or in the computer name
        retrieved by the 'hostname' command if /etc/hosts
        does not have a corresponding entry.

 (char(64))

OmsHeapBlockSize

default: 10000

change: OFFLINE

  each time the OMS Heap is growing, a memory block of OmsHeapBlockSize KB
  is required from the operating system

  The lower and upper limits are:
     10 <= OmsHeapBlockSize <= 524000

 (4 bytes integer)

OmsHeapThreshold

default: 100

change: OFFLINE

  defines the heap usage (in percent) when the livecache starts to
  reduce the heap usage of OMS versions, i.e. at transaction end all
  unchanged objects are removed from the version context.

 (4 bytes integer)

OmsSubHeaps

default: 1

change: OFFLINE

  defines the number of sub heaps of the OMS heap. In case of a single
  OMS heap becomes a bottleneck in multi CPU environments, multiple sup heaps
  may increase speed and scalability.

  The lower limit is:
    OmsSubHeaps >= 1 
     
  The upper limits are:
    OmsSubHeaps <= 128 and
    OmsSubHeaps <= MaxCPUs

 (4 bytes integer)

OmsVersionThreshold

default: 2097152

change: OFFLINE

  defines the size of a OMS version (in KB) when the liveCache starts to
  reduce the heap usage of that versions, i.e. at transaction end all
  unchanged objects are removed from the version context.

 (4 bytes integer)

OracleConcatNULLYieldsNULL

default: "YES"

possible values: "YES" "NO"

change: RUNNING

 OracleConcatNULLYieldsNULL 'YES' or 'NO'

 'YES': Concatenating a NULL value with a string yields a NULL result 

 'NO' : Concatenating a NULL value with a string yields a string result

PreloadObjectsDuringRedo

default: "NO"

possible values: "YES" "NO"

change: OFFLINE

new in 7.9.10.03

 This parameter specifies whether liveCache object containers are prefetched during redo (including standby databases).
 
 PreloadObjectsDuringRedo = 'YES' or 'NO'

 'YES': object containers are prefetched
 'NO' : object containers are not prefetched


RequestTimeout

default: 3600

change: RUNNING

 Restricts for all database sessions the waiting time for a lock to be released by other users.
 The time is specified in seconds.

 If a lock request cannot be satisfied within the time defined,
 a message is returned to the waiting database session.

 The lower and upper limits are:
      5 <= RequestTimeout <= 100000

 (4 bytes integer)

RequestUpdateStatisticsThreshold

default: 30

change: RUNNING

    If the optimizer detects that a table's size has changed by more
    than RequestUpdateStatisticsThreshold percent since the latest
    collection of statistics information, it will request the
    statistics information for this table to be updated by inserting
    a row with this table's name into the system table SYSUPDSTATWANTED.

RowLockManagementStripes

default: 8

change: OFFLINE

 Number of autonomous units in which lock collisions in rows are checked in parallel

 The lower and upper limits are:
     1 <= RowLockManagementStripes <= 64

 (2 bytes integer)

RowLocksPerTransactionThreshold

default: 50

change: RUNNING

If a transaction holds as many row locks as defined by this parameter, all further attempt to get a lock results in an escalation attempt.
The allowed range is 0 to 100 % and the default is 50 % of MaxSQLLocks.
If 0 % is set this additional escalation behaviour is disabled.

RowLocksToTableLockThreshold

default: 20

change: RUNNING

new in 7.9.10.00

If a table holds as many row locks as defined by this parameter, all further attempt to get a lock results in an escalation attempt.
The allowed range is 1 to 100 % and the default is 20 % of MaxSQLLocks. 0 disables lock escalation due to table locks at all.

SequenceCacheSize

default: 10

change: OFFLINE

 The value depends on the number of sequences used simultaneously.

 Use the database monitoring to obtain information about the
 sequence cache hit rate

 The lower and upper limits are:
     1 <= SequenceCacheSize <= 2147483640

 (4 bytes integer)

SessionTimeout

default: 900

change: RUNNING

 Maximum time of inactivity allowed for all database sessions.
 The time is specified in seconds.

 If no SQL statement is issued within the specified time, the database
 session concerned is terminated (ROLLBACK WORK RELEASE).

 The lower and upper limits are:
     30 < SessionTimeout <= 32400

 SessionTimeout = 0 : no timeout of session

 (4 bytes integer)

SharedSQLCleanupThreshold

default: 75

change: RUNNING

  Defines the threshold (%) from which on clean up occurs within SharedSQL

SharedSQLCommandCacheSize

default: 262144

change: RUNNING

 Specifies the maximum size (in KB) for SharedSQL statement caching (4 bytes integer).
 The lower and upper limits:
     32768 (=32MByte) <= UseSharedSQL_COMMANDCACHESIZE < 8388608 (=8GByte)
 In any case the initial size will be 16MByte, the supplemental size 2MByte.     

TableLockManagementStripes

default: 8

change: OFFLINE

 TableLockManagementStripes help to improve simultaneity when checking lock collisions
 on tables.
 The value of TableLockManagementStripes should be increased when the number of tables
 increases.

 The lower and upper limits are:
     1 <= TableLockManagementStripes <= 64

 (2 bytes integer)

TaskSchedulerStartupTimeout

default: 300

change: OFFLINE

new in 7.9.10.00

 Maximum time the TaskScheduler is allowed to spend in the startup-phase 
 The time is specified in seconds.

 The lower and upper limits are:
     30 < TaskSchedulerStartupTimeout <= 32400


 (4 bytes integer)

TraceBufferSizeALRTask

default: 5

change: OFFLINE

new in 7.9.09.01

 Size (in pages) of the memory used for kernel trace of task(s) of type ALR (automatic-log-backup-reader)

 The file is written cyclically.

 The lower limit is:
     0 <= TraceBufferSizeALRTask

 (2 bytes integer)

TraceBufferSizeALWTask

default: 5

change: OFFLINE

new in 7.9.09.01

 Size (in pages) of the memory used for kernel trace of task(s) of type ALW (automatic-log-backup-writer)

 The file is written cyclically.

 The lower limit is:
     0 <= TraceBufferSizeALWTask

 (2 bytes integer)

TraceBufferSizeFSTask

default: 5

change: OFFLINE

 Size (in pages) of the memory used for kernel trace of task(s) of type EV (event) / AN (dbanalyzer)

 The file is written cyclically.

 The lower limit is:
     0 <= TraceBufferSizeFSTask

 (2 bytes integer)

TraceBufferSizeGCTask

default: 20

change: OFFLINE

 Size (in pages) of the memory used for kernel trace of task(s) of type GC (garbage collector)

 The file is written cyclically.

 The lower limit is:
     0 <= TraceBufferSizeGCTask

 (2 bytes integer)

TraceBufferSizeIOCTask

default: 5

change: OFFLINE

new in 7.9.09.01

 Size (in pages) of the memory used for kernel trace of task(s) of type IOC (I/O-completion)

 The file is written cyclically.

 The lower limit is:
     0 <= TraceBufferSizeIOCTask

 (2 bytes integer)

TraceBufferSizeLWTask

default: 5

change: OFFLINE

 Size (in pages) of the memory used for kernel trace of task(s) of type AL (archive log)

 The file is written cyclically.

 The lower limit is:
     0 <= TraceBufferSizeLWTask

 (2 bytes integer)

TraceBufferSizeRRTask

default: 5

change: OFFLINE

new in 7.9.09.01

 Size (in pages) of the memory used for kernel trace of task(s) of type RR (redo-reader)

 The file is written cyclically.

 The lower limit is:
     0 <= TraceBufferSizeRRTask

 (2 bytes integer)

TraceBufferSizeSVPTask

default: 5

change: OFFLINE

new in 7.9.09.01

 Size (in pages) of the memory used for kernel trace of task(s) of type SVP (savepoint)

 The file is written cyclically.

 The lower limit is:
     0 <= TraceBufferSizeSVPTask

 (2 bytes integer)

TraceBufferSizeSVTask

default: 5

change: OFFLINE

 Size (in pages) of the memory used for kernel trace of task(s) of type SV (server)

 The file is written cyclically.

 The lower limit is:
     0 <= TraceBufferSizeSVTask

 (2 bytes integer)

TraceBufferSizeTITask

default: 2

change: OFFLINE

 Size (in pages) of the memory used for kernel trace of task(s) of type TI (timeout)

 The file is written cyclically.

 The lower limit is:
     0 <= TraceBufferSizeTITask

 (2 bytes integer)

TraceBufferSizeUSTask

default: 10

change: OFFLINE

 Size (in pages) of the memory used for kernel trace of task(s) of type US (user)

 The file is written cyclically.

 The lower limit is:
     0 <= TraceBufferSizeUSTask

 (2 bytes integer)

TraceBufferSizeUTTask

default: 5

change: OFFLINE

 Size (in pages) of the memory used for kernel trace of task(s) of type UT (utility)

 The file is written cyclically.

 The lower limit is:
     0 <= TraceBufferSizeUTTask

 (2 bytes integer)

TransactionHistorySize

default: 0

change: OFFLINE

 Information for long running transactions is kept within a wrap around buffer
 in memory. TransactionHistorySize defines the maximum number of entries hold within this
 buffer.

 If TransactionHistorySize = 0 then no information for long running transaction will be kept.


TransactionHistoryThreshold

default: 60

change: RUNNING

 If a transaction lived longer than TransactionHistoryThreshold seconds, then its
 will be inserted into the history buffer for long running transactions.


TransactionLockManagementStripes

default: 8

change: OFFLINE

 The specified number of regions allows for simultaneous transaction management.

 The value of TransactionLockManagementStripes should be increased when the number of
 parallel working sessions increases.

 The lower and upper limits are:
     8 <= TransactionLockManagementStripes <= 1024

     Note: values < 8 can be set in restart error situation
     When the database is online switch the value back to >= 8

 Default calculation, can be overridden
     MaxCPUs = 1 ---> TransactionLockManagementStripes = 8

 (2 bytes integer)

UpdateStatSampleAlgorithm

default: 1

change: RUNNING

    Selects the algorithm used for sampling during processing of 'Update 
    Statistics ... Estimate ...' statements.

    0: Use the classic algorithm (should not be used anymore) 
    1: Use an improved algorithm (Default setting)
    2: Use another improved algorithm that might have a greater tendency to
       overestimate the distinct value count.
    3: Use the classic algorithm with a different sampling algorithm.
    4: Use a combination of 1 and 2.


UpdateStatTempFiles

default: 0

change: RUNNING

new in 7.9.09.05

    Defines the number of temporary files used by 'Update Statistics' to count distinct values.  
    Increase UpdateStatTempFiles, if you observe huge numbers of collisions on 
    temporay files used by 'Update Statistics'.   
     If this parameter is set to 0, a value is chosen by  the kernel.

UseAlterTableAddColOptimization

default: "YES"

possible values: "YES" "NO"

change: RUNNING

If set to 'YES' default values of added columns are not materialized immediately. This saves execution time in case of large tables. 
If set to 'NO' all table rows are updated with the default values of the added columns.

UseDataCacheScanOptimization

default: "NO"

possible values: "YES" "NO"

change: RUNNING

 Specification of scan performance in the data cache

 'YES' : The complete data cache will be used for scans.
         This setting is advantageous if small tables are to
         be kept completely in the data cache.
 'NO'  : Only part of the data cache will be used for scans.

 (char(8))

UseHashedResultset

default: "YES"

possible values: "YES" "NO"

change: RUNNING

 UseHashedResultset 'YES' or 'NO'

 'YES': Hashed resultsets are used for aggregates
 'NO' : Hashed resultsets will not be used

UseIndexKeyColumnCompression

default: "NO"

possible values: "YES" "NO"

change: RUNNING

If set to 'YES' keys stored in indexes are stored in a compressed format. This saves disk space and allows more index data inside the data cache. On the other hand decompression needs some additional CPU power.

UseLobClustering

default: "YES"

possible values: "YES" "NO"

change: RUNNING

 'YES': Enable clustered storage of LOB file

UseParameterInfoForColumn

default: "YES"

possible values: "YES" "NO"

change: RUNNING

 UseParameterInfoForColumn 'YES' or 'NO'

 'YES': Application info description will be evaluated for native Column Parameter
 'NO' : Application info description will not be evaluated for native Column Parameter

UseQueryRewriteInDBProcedures

default: "YES"

possible values: "NO" "YES"

change: RUNNING

 UseQueryRewriteInDBProcedures 'NO' or 'Yes'

 'NO'  : Query rewrite won't be used at compile time of database procedures
 'YES' : Query rewrite will be used at compile time of database procedures


UseSharedSQL

default: "YES"

possible values: "YES" "NO"

change: RUNNING

 UseSharedSQL 'YES' or 'NO'

 'YES': SharedSQL is used for caching SQLCommands
 'NO' : SharedSQL will not be used

UseUnicodeColumnCompression

default: "NO"

possible values: "YES" "NO"

change: RUNNING

If set to 'YES' non-key columns of type '(VAR)CHAR UNICODE' and variable width
will not be stored in UCS2 but in UTF8 format on disk for newly created tables.
This will result in disk space savings at the cost of increased CPU
consumption.
Tables for which this optimization is enabled will show the value 'YES' for
column 'PACKEDUNICODECOLUMNS' in the system table 'FILES'.

Changing this parameter will not affect existing tables.

VolumeFormattingMode

default: "PARALLEL"

possible values: "PARALLEL" "SERIAL" "AUTO" (values depend on operating system)

change: OFFLINE

&if $OS in [ WIN32 ]
 VolumeFormattingMode 'PARALLEL' or 'SERIAL'

 'PARALLEL': Volumes are being formatted in parallel

 'SERIAL': Volumes are being formatted in serial
&else
 VolumeFormattingMode 'PARALLEL', 'SERIAL, or 'AUTO'

 'PARALLEL': Volumes will be formatted in parallel

 'SERIAL': Volumes will be formatted serially one after another

 'AUTO' : The database system forms groups of
          volumes which reside on the same device. Those groups
          will be formatted in parallel, but the members of the
          groups itself will be formatted serially.
&endif

 It is recommended that each volume is located on its own device.
 This way the maximum I/O performance for the database system is guarantied
 and parallel formatting achieves the best performance.

 Notice that this parameter setting only affects the format process 
 during initialization of the database (not used during ADD volume).

 The default setting is 'PARALLEL'.
 (char(8))