Why You Need It and How to Gain the Most Value From It
When terrorists crashed an airplane into the Pentagon September 11, it resulted in the deaths and serious injuries of about 75 percent of the people assigned to perform an accounting function for one of the U.S. military services. "It literally wiped out every single record they had for the entire year," recounts Kenn Kelley, senior systems analyst at Teksouth Corporation. "They had to rebuild the whole year, and they had 19 days to do it with only 19 percent of the people."
"Although Teksouth is not supporting that particular office," continues Kelley, "we have designed one of two data warehouse concepts the Services are staging against one another. The way our data warehouse is structured, we are capturing individual transactions on a daily basis; so we could have rebuilt that year of data for them literally overnight."
One of Teksouth's data warehousing solutions for a rival military branch illustrates the deliverables. It receives data feeds from more than 100 sites, with data files ranging from eight to 400 megabytes in size from each location. Just managing the file transfer process - let alone the data cleansing and data staging process in Teksouth's front-end processor - and then publishing the output reports is a gigantic feat. But the Alabama-headquartered high-tech firm was able to execute queries to run against the database and produce output files with a global data refresh every 35-45 minutes throughout the fiscal year closeout on September 30. "Even Microsoft wants to know how we are able to do that!" says Kelley. "The most data that would be lost would be anything that was done during the current day."
Comparing Apples to Oranges
On anybody's list of business solutions (for the unique, dynamic changes wrought by the driving force of the Internet) are tools that reduce time to market and achieve flexibility as well as multi-dimensional performance. Even in an economic slowdown, when organizations are forced to control their information technology spending sprees, a data warehouse is a critical strategic component of enterprise-wide initiatives. It exploits data - an organization's most valuable asset - which accelerates the decision-making process. There's no more valuable resource in a rapidly changing marketplace than timely, relevant information.
The ultimate value of a data warehouse is not as a simple repository for transactional data and generation of reports but, rather, in its ability to process complex, disparate nuggets of data, then slice and dice them into any number of meaningful views for analysis.
"People think you can't compare apples to oranges," says Kelley. "But you can compare the color and thickness of the skin, the structure of the fruit, how the seeds are stored, the growing conditions, or even compare them relative to a quince or a lemon."
A data warehouse gives an organization the ability to pull in information from all of its different business units or communities - with all of their silo accountability and transactional systems - and come up with ways to compare, contrast and manipulate the data for meaningful analysis.
And that is really the strength of a data warehouse. Organizations implementing this business strategy need to be farsighted enough to include data that is even somewhat questionable as to its importance. Something that may not be important today could be tremendously important down the road. The strategy going in should not be simply to mirror functions and operations as they currently are, but to allow for expanded use of information.
Return on Investment
Some of the healthcare industry's greatest advances in patient care are based on insights gained through use of data warehouses. Cardiac surgeons, for instance, can access extensive real-time information on which to base their diagnoses, more accurately assess patient risk and reduce errors by accessing data residing in the Society of Thoracic Surgeons' data warehouse. Insurance underwriters use data warehouses to analyze and predict risk factors for particular customer characteristics, thus reducing costs and liability.
The airline industry, whose profitability now lies in ruins, must seek ways to reduce costs. Systems analyst Kelley says a data warehouse is a means of capturing costs associated with flying airplanes and then analyzing the data to design cost-cutting solutions involving just-in-time maintenance, just-in-time inventory and just-in-time logistics.
Teksouth designed, hosts and maintains a unique data warehouse for one of the U.S. Military Services. It's the first of its kind, in that it integrates multiple disciplines and multiple functional areas from the financial, logistics, operations, training and payroll operations. It tracks data on the real-time status of high-ticket military items - their whereabouts, maintenance status, specialized components and mission profile. It enables the Service with a particular need at a particular location to determine the real-time location and status of the exact item needed, versus the time it will take to modify an item already at the needed location.
Citing another valuable function of a data warehouse, Kelley says companies are experiencing a capital intelligence loss because people who understand all the impacts of an individual transaction are retiring. "You might think a transaction is a sale. But it's really a reduction in inventory and an increase in demand within the supply chain," he explains. "The people who have been doing this for a number of years can do it in their heads, whereas the new folks who are going to be responsible for it don't have a clue. A data warehouse captures corporate intelligence and converts it into business rules that define the impacts of any given action or function."
Not All Data Warehouses are the Same
What are the critical elements to consider in choosing a data warehousing outsourcer and implementation best practices that ensure success? First and foremost are security concerns - what measures has the service provider put in place to protect an organization's data (especially where the warehouse is offsite and the client users access information through a network)?
Chris Somers, Teksouth's technical director, says his company solves the security issues with an "access and analysis system" - a separate server implementing all sorts of security measures. "Our warehouse databases are not accessible to anyone - not even the customer's users. It's completely isolated," says Somers. "The access system is the one that has the query account. Users are authenticated on that system, and then it will query the warehouse databases on the user's behalf."
To capture more value, it's also important to select a service provider whose data warehouse is designed to accommodate a number of varying types of data inputs (i.e., DOS database, relational database, text database feeds, and spreadsheets).
Of equal importance is the provider's capability to provide a wide and flexible range of functionality in the output analysis from the data. Teksouth customizes this functionality to its customers' needs, with access either through Web pages or through the use of a client-server-based Online Analytical Processing (OLAP) query tool. (OLAP technology enables end users to analyze data and make decisions.)
Teksouth automatically gives its customers the ability to pull their own information and build their own query designs. Kelley says, "We've seen that things change on a day-to-day basis, and a customer never knows what is going to happen on any given point in time."
"The key is to remain flexible," he continues. "We train the users and make sure they have sufficient information on the use of the extract and display so that they can roll with the punches and be able to turn 90 degrees and charge off in a different direction. The data elements that are chosen are up to the customer. They can provide filtering either in the design stage or on-the-fly within a particular data field. Being able to do that without returning to the design screen and generating a new cube file is a tremendous advantage. It gives the customer the ability to produce unlimited numbers of output or display from one output."
Implementation, according to Somers, can be as short as three months for a small or mid-size warehouse or in a year and a half for a larger one. The primary challenge encountered during implementation, he says, is the customer's ability to properly define its expectations as to what it wants the data warehouse to provide. Clearly stated upfront objectives will shorten implementation time and ensure success. Choose a service provider that will perform a 'big picture assessment' of organizational needs and processes to discern what data needs to be included to achieve stated customer goals.
As business goals change (affecting data queries), geographic markets and offices are acquired or divested (affecting data input), the flexibility of a data warehouse becomes even more valuable. "All data goes into the warehouse pretty much as you collect it - what's there is the record of your business," says Somers. Once it's implemented, the customer simply pays for maintenance of the reports and updating the data according to the parameters the customer has defined. Customers have flexibility to generate changes in reports to fit their changing business needs.
Validating customer data is another key component for success. Somers explains the Teksouth methodology: "We do an initial functional review and then a prototype implementation of the data. We find it helps to have a customer representative there to validate data during implementation. Even in production mode, we continue to maintain data integrity. Each daily load compiles reconciliation information in a separate entity and balances it against the entities that users query. If it doesn't balance, it puts out an exception, which is then sent back to the customer for validation."
There's no one-size-fits-all approach to data warehousing. Thus, it's critical to choose a service provider with expertise in all of the complexities and aspects involved - architecture, systems, databases, business processes and functions. For a data warehouse solution to achieve maximum value, it must be approached as a long-term strategy. So look for a partner that not only will be responsive to its customers' requirements but also flexible and proactive in working to achieve long-term objectives.
Lessons from the Outsourcing Journal:
- Because of the complexity of data warehouse solutions, it's best to choose a single-source provider with service capabilities for designing, implementing, integrating and hosting the entire process.
- Choose a provider offering broad functionality in data input and query output.
- Be sure to include data that is somewhat questionable as to its importance, as it may become more important down the road.