Setting Up Direct Insertion to Collect Information for Usage Tracking

9-6 System Administrators Guide for Oracle Business Intelligence Enterprise Edition The subsequent integer columns are valid only if the Success indicator column 12 signifies a successful query value is 0: ■ Timestamp columns. The Start timestamp and End timestamp columns indicate the wall clock time local time when the logical query started and finished. Each value is 19 bytes of character data that represents a SQL-92 timestamp. The format is yyyy-mm-dd-hh:mm:ss. The related columns, Start date and End date, contain just the date component from the respective timestamps in the yyyy-mm-dd format. Finally, the related columns, Start hourMin and End hourMin, contain just the hour and minute components from the respective timestamps in a char hh:mm format.

9.1.2.4 Performance Considerations

When usage tracking is enabled, the BI Server collects usage tracking data for every query. This data, however, is written to disk only at user-specified intervals, known as checkpoints. The default setting is to have checkpoints every 5 minutes. Although you can modify this value in the NQSConfig.INI file, reducing the interval adds overhead and, if set low enough, could potentially impact server performance. Setting the value to a larger number increases the amount of usage tracking data that could be lost in the unlikely event of an abnormal shutdown of the BI Server. The BI Server periodically initiates usage tracking log file rollovers. A rollover consists of closing the current usage tracking log file and opening a newly created one for writing subsequent data. The frequency at which rollovers occur is called a rollover interval. The default rollover interval is 240 minutes every 4 hours. Usage tracking log files that are closed are available for analysis. Setting a lower rollover interval makes usage tracking log files available for analysis sooner, but at the cost of additional overhead. If the checkpoint interval equals or exceeds the rollover interval, then only the rollover occurs explicitly. The checkpoint occurs implicitly only when the old usage tracking log file is closed.

9.2 Description of the Usage Tracking Data

Table 9–2 describes each column in the usage tracking table. Where appropriate, the data type and length is also included. Table 9–2 Usage Tracking Data Column Description CACHE_IND_FLG Default is N. Y indicates a cache hit for the query; N indicates a cache miss. COMPILE_TIME_SEC The time in seconds that is required to compile the query. CUM_DB_TIME_SEC The total amount of time in seconds that the BI Server waited for back-end physical databases on behalf of a logical query. CUM_NUM_DB_ROW The total number of rows that are returned by the back-end databases. END_DT The date the logical query was completed. END_HOUR_MIN The hour and minute the logical query was completed. Managing Usage Tracking 9-7 END_TS The date and time the logical query finished. The start and end timestamps also reflect any time that the query spent waiting for resources to become available. Note: If the user submitting the query navigates away from the page before the query finishes, then the final fetch never happens, and a timeout value of 3600 is recorded. However, if the user navigates back to the page before the timeout, then the fetch completes at that time, and this is recorded as the end_ts time. ERROR_TEXT Default is Null. Varchar250 Error message from the back-end database. This column is only applicable if the SUCCESS_FLG for more information, see entry later in this table is set to a value other than 0 zero. Multiple messages are concatenated and are not parsed by the BI Server. NODE_ID The host name of the computer where the BI Server is running. NUM_CACHE_HITS Default is Null. Number10,0. For DB2, the data type and length is Decimal10,0. Indicates the number of times that the cache result returned for query. NUM_CACHE_ INSERTED Default is Null. Number10,0. For DB2, the data type and length is Decimal10,0. Indicates the number of times that the query generated a cache entry. NUM_DB_QUERY The number of queries that were submitted to back-end databases to satisfy the logical query request. For successful queries SuccessFlag = 0 this number is 1 or greater. PRESENTATION_NAME Default is Null. Varchar128 The name of the Oracle BI Presentation Catalog. QUERY_BLOB The data type is ntext when using SQLServer, and is CLOB when using ORACLE, DB2, or TERRADATA databases. Contains the entire logical SQL statement without any truncation. QUERY_KEY Default is Null. Varchar128. An MD5 hash key that is generated by Oracle Business Intelligence from the logical SQL statement. QUERY_SRC_CD The source of the request for example, Drill or Report. QUERY_TEXT Varchar 1024. The SQL statement that was submitted for the query. You can change the length of this column using the ALTER TABLE command, but note that the text that is written into this column is always truncated to the size that is defined in the physical layer. It is the responsibility of the repository administrator not to set the length of this column to a value greater than the maximum query length that is supported by the back-end physical database. For example, Oracle Databases enable a maximum Varchar of 4000, but Oracle Databases truncate to 4000 bytes, not 4000 characters. Hence, if you use a multibyte character set, the actual maximum string size has a varying number of characters, depending on the character set and characters used. REPOSITORY_NAME The name of the repository that the query accesses. Table 9–2 Cont. Usage Tracking Data Column Description