|
SNOWGEN creates snowflake tables for a star schema.
Example
c:\omnidex\demo>snowgen
DS73000 1.0.99 SNOWGEN - Snowflake Generator Thu Mar 18 12:57:58 2004
(c) Dynamic Information Systems Corp. 1981-2004
LICENSEE: ** Trial Copy **
Snowflake tables can be generated using two approaches:
1. Use OmniAccess to read unsorted data, and then create up
to 32
snowflake tables by writing to in-memory hash tables, eliminating
duplicates, performing in-memory sorts, and then writing output tables
with an appropriately sized integer code field. This is the default
approach, and the best performing approach, especially since multiple
snowflake tables can be produced in one pass, but it also requires
sufficient core memory.
2. Use OmniAccess to export sorted data into a temporary table,
eliminating duplicates, and then write a single output table with an
appropriately sized integer code field. This approach requires little
memory, but is I/O intensive, as a temp file is produced by OmniAccess
to sort the data, plus an intermediate table is used while eliminating
duplicates and determining the appropriate size for the integer code
field. Furthermore, only one snowflake table may be created per pass
of the data, since OmniAccess is used to return data in sorted order.
Approach to use [1]:
Approach to use [1]:
Environment file name:
Dimension table name:
Snowflake table #1
Column in dimension table to turn into snowflake table:
Additional columns to include in snowflake table:
Output filename for snowflake table:
Estimated cardinality of snowflake table (only needed in approach 1):
Top
|
