|Bookshelf Home | Contents | Index | PDF|
The parameters in the Cache section provide configuration information for Siebel Analytics Server caching. For information about caching, see the chapter on query caching in Siebel Analytics Server Administration Guide.
Specifies the directory paths for where the cached data is stored and the maximum capacity in bytes, kilobytes, megabytes or gigabytes. The maximum capacity for each path is 4 GB. For optimal performance, the directories specified should be on high performance storage systems.
Each directory listed needs to be an existing, fully-qualified, writable directory pathname, with double quotes ( " ) surrounding the pathname. Specify mapped directories only. UNC path names and network mapped drives are allowed only if the service runs under a qualified user account. To change the account under which the service is running, see the topic Changing the Account Under Which a Windows Service Runs.
Specify multiple directories with a comma separated list. When you specify more than one directory, they should reside on different physical drives. (If you have multiple cache directory paths that all resolve to the same physical disk, both available and used space may be double-counted.)
NOTE: Specifying more than one directory per drive does not improve performance, because file input and output (I/O) takes place through the same I/O controller. In general, specify only one directory per disk drive. Specifying multiple directories on different drives may improve the overall I/O throughput of the Siebel Analytics Server internally by distributing I/O across multiple devices.
The disk space requirement for the cached data depends on the number of queries that produce cached entries and the size of the result sets for those queries. The query result set size is calculated as row size (or the sum of the maximum lengths of all columns in the result set) times the result set cardinality (that is, the number of rows in the result set). The expected maximum should be the guideline for the space needed.
This calculation gives the high-end estimate, not the average size of all records in the cached result set. Therefore, if a result set's size is dominated by variable length character strings and if those strings' lengths have a normal distribution, you would expect the average record size to be about half of the maximum record size.
Specifies the size of the buffer pool to use to store information from the cache metadata file in memory to improve the cache response time. Specify KB for kilobytes, MB for megabytes, and no units for bytes.
Specifies the maximum number of cache entries allowed in the query cache. When the limit is reached, old entries are replaced by new entries according to the mechanism specified by the REPLACE_ALGORITHM parameter. The actual limit of cache entries might vary slightly depending on the number of concurrent queries. The default value is 1000.
A Siebel Analytics user may have a cached result set containing information at a particular level of detail (for example, sales revenue by ZIP Code). A second query may ask for this same information, but at a higher level of detail (for example, sales revenue by state). Siebel Analytics can satisfy this second query by aggregating data from the first result set stored in the cache. That is, Siebel Analytics sales revenue for all ZIP Codes in a particular state can be added to obtain the sales revenue by state. This is referred to as a rollup cache hit.
Normally, a new cache entry is not created for queries that result in cache hits. You can override this behavior specifically for cache rollup hits by setting POPULATE_AGGREGATE_ROLLUP_HITS to YES. Nonrollup cache hits are not affected by this flag. If a query result is satisfied by the cache—that is, the query gets a cache hit—then this query is not added to the cache. When this parameter is set to YES, then when a query gets an aggregate rollup hit (for example, "sales by region" is answered from "sales by district, region") then the result is put into the cache. Setting this parameter to TRUE may result in better performance, but results in more entries being added to the cache.
Specifies the amount of time, in minutes, between each backup of the cache metadata file. When caching is enabled, the cache metadata file is always backed up upon server startup and is backed up again at the interval specified in the parameter. The server automatically uses the backup file after a sudden server shutdown occurs (for example, a power failure or a crash). The backup file is created in the same directory as the cache metadata file and is named the same as the cache metadata file with a .bak extension appended to the name.
When the backup file is used, any entries cached after the backup file was created are automatically purged from the cache. If any entries have been purged or aged out of the cache after the backup file was created, those entries are not recovered. To force the creation of a new cache metadata backup file, cleanly shut down and restart the server as described in the chapter on administering the query environment in Siebel Analytics Server Administration Guide.
|Siebel Analytics Platform Installation and Configuration Guide|