Real World Problem #4 "Corp. Data"
Data Warehousing & Data Mining
Imagine for a moment a computing wonderland. Where the least expensive
solution is the fastest and the most reliable. A land where any
level of performance is possible. Where doubling your capacity
means doubling your cost, period. A land where development time
is measured in hours and days, instead of months and years. Where
first prototypes can safely be used in full scale production.
A land where change is painless and controlled. Where complete
code testing is not only possible but practical.
How is this possible? By avoiding many of the common DBMS problems.
- No need for meta-data. It contains less
information than the original.
- No data normalization. Normalized data
is harder to mine.
- No conventional DBMS. There is no operating
system on top of your operating system.
- Scale predictably, with a better
than linear cost/performance curve.
- Eliminate barriers to growth with no
architectural upper limits.
- Save money,
cost savings between 10 to 1 and 100 to 1 are possible.
Other than avoiding problems, what will MPbase add to the
picture?
- Increase reliability by using a paradigm
that allows inexpensive parallel redundant systems. 100% uptime
can be both practical and cost-effective.
- The speed to strip mine your raw data; conventional
data mining is like taking core samples and working with averages.
- Demystify your data by
storing it in a more natural format, one that highlights errors
and discrepancies for easy identification and correction.
- Gain controllability with a paradigm that
allows exhaustive testing and parallels.
How about access?
- Easy powerful interfaces to Visual
Basic, HTML, PERL, C, UNIX command line. Almost any system can
directly access a MPbase.
- Put the data on your intra-net to allow
any system you authorize access to the data.