Data Warehousing
Data Warehousing and Decision Support System (DSS) systems are finally becoming less of a "fad" and more of a
necessity for businesses around the world. Data Warehousing and DSS are NOT the
same thing, a Data Warehouse is where data is stored to support DSS. DSS is
the access and analytical representation of data stored in a Data Warehouse.
Information must be retrieved quickly and represented in a readable
format for users, the information must be correct and readily available.
Companies are recognizing that the sooner they get their information, the
quicker they can react to changing market needs - No longer can they wait
for a monthly or weekly report.
Why Tandem?
Tandem hardware is capable of storing 5,164 TB (Terabytes) of data using 18GB
disk drives (or 5.04 PetaBytes) and connecting 4,080 processors executing
parallel paths to the data. Tandem hardware is continuously available
(99.99% availability). The Himalaya line of servers is one of the most powerful
available, and with Tandem's unique ServerNet architecture they boast some of
the fastest data transfer rates in the industry - including holding the world
record benchmark for TPC-C and TPC-D for DSS & OLTP.
What makes DSS difficult?
The perception about building a Data Warehouse is that it is simple to accomplish, after all, "The
data exists on my legacy system - Just copy it to the warehouse and install ODBC
on my users workstations - Right?" - Well, unfortunately it's not always that
simple. Transferring data from other systems such as IBM, DEC, and HP requires
sophisticated data loading techniques. EBCDIC conversion, date/time conversions,
averaging, and joining data from other tables requires the use of tools such
as DataBuild. Speed is also an issue, when loading millions of records in a
limited time window, inserting one record at a time into the database just
doesn't cut the mustard. Data needs to be copied in bulk transfer modes, in
parallel threads, using buffered file access. Once data exists in the Warehouse
then the maintenance work begins, managing disk space, TMF dumps, validation,
indexes, views, and report programs. The typical IS staff is lacking the
information and experience to fully investigate and perform these tasks.
Why do I need a Consultant?
Building a Data Warehouse requires a well thought-out plan, an understanding
of the system architecture, a comprehensive database model, and access to tools
and utilities that are not widely known. It is difficult to find a member of your IS
staff that has performed even one Data Warehouse dataload. In large data loading,
mistakes can be costly. If it takes 5 hours of processing time to load 5
million records, and the results are incorrect, you have just lost 5 hours of
processing (assuming you caught the error on the first day of processing). A
Data Warehouse Consultant from Randall Consulting, that has many years of
experience performing data loads knows the ropes and can minimize mistakes,
saving many hours of processing (and re-processing) time.
(Return to top)
Send Comments and Suggestions to: Webmaster
