Possible config entries for RDB archiving

If alarms where the pre-alarm is included in an already exported archive set occur, the database tries to reach the old alarm for five minutes. Thereby, alarms are possibly held back and the alarms are not written to the database. To prevent this and to reduce this time e.g. to 30 seconds use [ValueArchiveRDB] "alertUpdateDelay = 30".

Specifies the storage mode, with which the data blocks are buffered, before they will be written into the database.

• bufferToDisk = 0

  Without BufferToDisk: in case of a disconnect to the RDB, the data blocks are buffered in RAM and accordingly they are written into the database.

  Note:

  Data will get lost in case no free memory is available
• bufferToDisk = 1

  BufferToDiskMin (default)

• bufferToDisk = 2

  BufferToDiskMax

Absolute path to the directory, in which the data blocks are buffered onto the HDD using bufferToDisk = 1|2.

If the default directory is used the directory buffer will be created automatically.

If you use an individual storage place, you have to enter the absolute path, e.g. "D:\BufferToDisk\Files".

Database password.

Timely delay of the RDB initializing process in seconds after database start.

If the database was stopped and is restarting again, it may happen that although the connection between RDB and the database has been established, the database start-up process has not been complemented yet. With this config entry RDB awaits the defined time before initialization of the connections and thus begins to write data into the database.

By default this config entry is set to 30 seconds.

In case of a database connection loss or if a connection will be opened/closed via the "closeDBConnection" and "openDBConnection" internal datapoint elements, this delay is taken into account.

In case of a RDB manager start this delay is not used.

Indicates the mode of database read queries:

• 0 = datapoint name or DPE name; although this option impacts on performance, it does allow you to access the data in the Oracle database from a different WinCC OA project.
• 1 = IDs (faster)

Number of threads (and also connections to the database) the RDB manager uses to (parallel) read (query) operations in the database.

When using queryRDBdirect = 1, the read connection for the direct DB access is not affected by this setting (maxRequestThreads can be set to 0 in this case).

RDB manager only opens a single write connection to the database, all other connections (update, deletion, info) will be opened (and closed afterwards) when necessary. Performance is worse in comparison to leaving all connections open all the time (openConnOnDemand = 0).

Number of read connections is not affected by this setting (use maxRequestThreads instead). When using queryRDBdirect = 1 the read connection for the direct DB access is not affected by this setting. The openConnOnDemand entry is required for large, distributed systems when several RDB managers write into the same database. Otherwise Oracle requires too much memory since too many connections are open simultaneously.

An additional DB connection is established for updates. Is only used if openConnOnDemand = 1 (see above) ,

This update connection is closed again updateConnCloseDelay seconds after its last use. Default = 180 (3 minutes).

Specifies whether the function dpGetPeriod() should query outside the specified period (the parameter "Count" of the function dpGetPeriod ) or not. If the parameter "count" of the function is bigger than 0, also values outside (this means before and after) the queried period are queried. This may sometimes take longer. queryOverBounds = 1 means that values are queried outside the specified period.

This works with queryRDBdirect = 1 only then, when one datapoint element is queried at the same time. queryOverBounds = 0 queries only values within the period. Thus, the query is faster.

Aborts database queries after <queryTimeout> seconds.

If you set the value to 0, the queries are not cancelled.

With the config entry queryFunction = 1 a database function is used for dpGetPeriod() instead of a query. Therefore, the limitations such as the number of tables and the length of the SQL statement cease to exist. With queryFunction = 0, a query is used as so far.

The entry can be used in ValueArchiveRDB and UI sections.

The Config entry allows to specify, if an archive group is mapped to an other archive group to prevent the creation of a new archive group and therefor improving the overall performance.

Example

redirectArcGroup = VA10 :QPS

The archive group VA10 is redirected to the archive group QPS

redirectArcGroup = VA* :EVENT

The archive groups with a name beginning with "VA" are redirected to the EVENT archive group.

• sendMaxTS = 1

  The RDB manager gets the highest time stamp from the database and synchronizes the data with the Data manager. Thus, the Data manager sends all value changes that are newer than the time stamp from the database, to the RDB manager.

• sendMaxTS = 0

  The RDB manager does not get the time stamp from the database and does not synchronize the data. This improves the performance but the last value change until the RDB start is not saved in the Oracle database.

• sendMaxTS = 2

  The latest time stamp per system is saved in the SYSTEMS table and updated with every block insert. When the RDB manager is started, the time stamp is read from the SYSTEMS table. This is a quick read operation that allows you to query the last time stamp of your system.

• 1 = Write data to RDB archives using OCI Oracle Call Interface (OCCI). To activate the OCI bulk writing, set this config entry to 1. This improves the performance.
• 0 = Do not use OCI.

Additionally, there are own APM query functions, which can be used to query another RDB database. The functions use the DPE name instead of the element ID for the query. For the APM query functions see chapter Direct read functions.

APMDbUser = User of the database.