Best Practices for Running the Web Server Log Import DTS Task

Article
11/12/2009

This topic lists best practices and important notes for running the Web server log import task.

If you have multiple applications and log files, use a script to configure the Web log import DTS instead of using the user interface. For instructions, see Scripting the Web Server Log Import DTS Task.
If you are running the Web log import DTS task by using a script, you must specify a 3 gigabyte (GB) switch in the Windows 2000 Boot.ini file on the computer hosting the Data Warehouse (SQL Server) and on the computer hosting the Analysis server. For instructions, see Running the Web Server Log Import DTS Task for a Large Site
The Web log file import process will not import log files while Internet Information Services (IIS) is actively writing to them. Do not import your log file directly from your Web site. Instead, verify that the server is no longer accessing the file, and then copy the file to a different directory. You can then import it from the directory you selected.
If you are importing Web log data for multiple sites, and you perform a Data Warehouse level operation to import data for all sites associated with the Data Warehouse, the Web server log import task imports all log files in the log path depending on the criteria set for All logs since last import, or Logs by date/time range. The Web log import task does not recognize that you might have cleared the selection for some log files at the site level that you do not want to import.
If you want to import a missed Web log file into the Data Warehouse, delete all of the Web logs imported after the day missed and then re-import the Web logs. Do not import the Web log files out of sequence; if you do, the open visits may be truncated prematurely.

Use the Data deletion DTS task to delete the imported data. To delete orphaned dimension data, use the DELETE ORPHANS clause, but note that it takes a long time to run. For more information, see Deleting Orphan Data from the Data Warehouse.
Strings that are stored as a global unique identifier (GUID) must begin and end with curly brackets “{” and “}” before they are written to the Web log file. This is the case with user_id, which is processed by the Profile data import task, and address_id, which is processed by the Transaction data import task.
If you are using the Campaign features of the Targeting System, the Web server log import task imports the URIs for campaign events, even if you do not run the Campaign data import task. The Web log files are the source of this data.
For information about importing log files from non-Commerce Web servers, see Knowledge Base article Q293887 HOWTO: Analyze Non-Commerce Web Server Log Files, located at https://go.microsoft.com/fwlink/?LinkId=5166.
Commerce Server only supports files in W3C extended log format. If you are importing Apache or Lotus logs, you must convert them to that format before you can import them.
If you want to import a Web log that was missed (out of sequence) into Commerce Server, delete all the Web logs that you imported after the missed day, and then re-import the missed Web log file.
Use the following performance counters for this task:
- UImport: Free hit buffers. The number of currently available empty hit buffers.
- UImport: Used hit buffers. The number of currently filled hit buffers.
- UImport: Hits per second. The rate per second at which the import task is processing hits.
- UImport: Current visits. The total number of active visits recorded while the log files are processed.
- UImport: Commerce events. The total number of Commerce Server events that occurred for the import task.
- UImport: Decoded cookies. The total number of Commerce Server-decoded cookies encountered while the import task was running.
- UImport: Query strings. The total number of query strings processed while the import task was running.

Best Practices for Running the Web Server Log Import DTS Task

See Also

Additional resources