VIEWGEN creates flattened data warehouse views of multiple tables to
facilitate high-speed retrieval.
The VIEWGEN program prompts for whether to use memory-mapped access or
malloc'd memory for its internal processing. Memory-mapped access may
allow access to more of the machine's memory on many platforms.
Performance can be degraded using mapped access, however on most platforms,
benchmarks showed that performance degradation was less than 5%. Note
that when using memory-mapped access, it may be necessary to direct each
dimension and snowflake table to a separate device, as a temporary file
must be created for each dimension table. Unless overridden, these files
are created in TMPDIR.
VIEWGEN also has added prompts for selection criteria for each table,
allowing for filtering of dimension, snowflake and fact tables. This can
be valuable to reduce the size of dimension tables loaded into memory.
These prompts can be avoided by setting the environment variable VIEWGEN_NOMAP=1.
This allows existing scripts to be run without change. This will be removed
in future releases.
VIEWGEN now produces a file containing an insertable environment file
entry for the superview table. This file is overridden each time VIEWGEN
is run, and is called "view.src". VIEWGEN also includes improved
statistics at the end of each run, showing the amount of time and memory
required.
Example
Following is a sample run of the new VIEWGEN program:
DS73000 2.0.00 VIEWGEN - Super-View Generator Sun Jun 06 23:22:56
1999
(c) Dynamic Information Systems Corp. 1981-1999
LICENSEE: ** D.I.S.C. Internal Copy **
Environment file name: star.env
Fact table name: orders
Columns to include in view: status, tax_state, source, pmt_method,
discount, quantity, sales_tax, amount,total
Options passed to oaselect:
Criteria passed to oaselect:
Estimated cardinality: 200
Process dimensions separately? n
Merge dimension result files? n
(M)apped access or m(a)llocs? m
Dimension table #1: prospects
Fact link columns: acct
Dimension link columns: acct
Columns to include in view: acct, state, zip, region, country,
gender, mailflag, mailing, profession, income, profile, preferred, frq_buy_mbr,
cr_limit, discount_rt, cust_since, last_order
Options passed to oaselect:
Criteria passed to oaselect:
Result file name (optional):
Scratch file name (optional):
Estimated cardinality: 10
Join type (I)nner/(O)uter: I
Snowflake table #1: /
Dimension table #2: products
Fact link columns: product_no
Dimension link columns: product_no
Columns to include in view: product_no, division, dept, category,
mfr
Options passed to oaselect:
Criteria passed to oaselect:
Result file name (optional):
Scratch file name (optional):
Estimated cardinality: 100
Join type (I)nner/(O)uter: I
Snowflake table #1: /
Dimension table #3: dates
Fact link columns: order_date
Dimension link columns: dt
Columns to include in view: dt, yr, mo, dy_of_mo, dy_of_yr,
dy_of_wk, wk, qtr, fqtr, fyr
Options passed to oaselect:
Criteria passed to oaselect:
Result file name (optional):
Scratch file name (optional):
Estimated cardinality: 4383
Join type (I)nner/(O)uter: I
Snowflake table #1: /
Dimension table #4: /
Rows to process (CR for all):
Number of output files: 1
Filename for view file 1: osv.dat
File 1 limit (# MB or 'none'): none
Print status messages? D1
svw.debug: 1
Creating super-view ...
Loading dimension table prospects ...
Loaded 10 rows from dimension table: prospects
Loading dimension table products ...
Loaded 100 rows from dimension table: products
Loading dimension table dates ...
Loaded 4,383 rows from dimension table: dates
Beginning scan of fact table orders
Processed 200 rows in 00:00:00
Wrote 200 rows to output file osv.dat
Rejected 0 rows due to inner join failures.
Super-view produced successfully.
Number of rows: 200
Inner join rejections: 0
Time required: 1.242 CPU sec. 3.295 Elapsed sec.
Load of dimensions: 1.202 CPU sec. 2.884 Elapsed sec.
Processing fact: 0.030 CPU sec. 0.190 Elapsed sec.
Fact rows per hour: 249,653
Number of columns: 41
View record length: 80
Number of files: 1
Disk space: 0.01 MB
Max malloc'd memory: 0.05 MB
Max mapped memory: 0.19 MB
Max combined memory: 0.25 MB
Max heap size: 0.29 MB
Top
|