How Tanker Works
Tanker writes multiple, filtered audit files simultaneously. This means that multiple DATABridge clients (each with their own filters) can replicate a DMSII database without DBEngine having to read the entire audit file for each client. Thus, I/O time on the host is reduced and overall throughput is increased, as shown in the following diagram:
Filtering refers to any filters you create for a tailored support library. Typically, you create filters to select certain records and data items for replication. See “DATABridge Support Library Filtering” in the DATABridge Programmer’s Reference for information on creating filters.
Suppose a bank has 30 branches, each with a DATABridge client replicating to an SQL Server database. Further, the bank creates a filter for each branch that selects the records that branch should receive.
Without Tanker, each time any of the 30 DATABridge clients ran, DBServer would request that DBEngine read the entire audit file. Then, DBServer would request that the Support Library specified in the DBServer parameter file filter the audit file. Then, the filtered data would be sent to the DATABridge client. This would be repeated for each DATABridge client.
With Tanker, however, DBEngine would read each audit file once and Tanker would create a filtered audit file for each branch with only that branch’s records. When the DATABridge client at the branch runs, DBServer uses the filtered audit file for that particular client. You can remove the original audit file any time after Tanker finishes reading it and starts creating the filtered audits for the next audit file.
DBServer and Span can both use filtered audit files. For either accessory to use the audit files filtered by Tanker, set the PREFILTERED and FILTER options in their respective parameter files.