|Sign-Up for Free Exclusive Services:||Portals|||||eNewsletters|||||Web Seminars|||||dataWarehouse.com|||||DM Review Magazine|
|Covering Business Intelligence, Integration & Analytics||Advanced Search|
Like many of us, I've been navigating the maze of Web sites, magazines, television shows and newsletters claiming to provide the "guide to investing in 2003" or "ways to beat a bear market." The majority of these articles try to lure me in by highlighting results of various investment options that have performed quite well over five and/or 10-year periods. For a moment, I imagine the end of all my financial concerns, opportunities with guaranteed returns that are void of risk. What a relief! But my enthusiasm wanes as I notice the glaring asterisk placed adjacent to the glorious figures. Skimming down to the bottom of the page, I see the footnote that states, "Past performance is no guarantee of future results." Or, in other words, "Good luck, but you're on your own from this point forward."
IT executives face similar concerns. How can you narrow the broad range of potential projects and focus on those that deliver maximum short-term and long-term value? The good news is an investment option does exist that provides results that can be forecasted with extreme accuracy and minimal risk. Think of it as a highly rated bond, but with returns in the 200-500 percent range! For those organizations looking to reduce operating costs and/or improve analytical capabilities, the answer to high return IT investing in 2003 is data mart consolidation (DMC).
IT managers continue to search for ways to reduce costs and quantify the return on their business intelligence dollars, yet very few data mart projects have been measured in terms of ROI. Those that have been measured generally prove that aggregate data mart costs have spiraled out of control. In fact, it is not unusual to see data mart support costs that exceed the intended benefits of the project! This problem is not readily apparent because the majority of data mart projects are funded at the department level. If these redundant departmental costs were collected across the enterprise, the sum would absolutely stun most executives.
According to Bob Parker of AMR Research, as stated in Five High-Value Infrastructure Projects for the 2003 Budget (September 2002), "The cost to maintain a data mart is between $1M and $2M." Yet surprisingly, "Between 35 and 70 percent of these costs are redundant across data marts."
Data marts may have been a reasonable approach to satisfy urgent user requests constrained by departmental budgets years ago; however, now this path of data mart proliferation has resulted in a never-ending list of mundane, unproductive tasks (and costs) required to coordinate and synchronize disparate marts. To be fair, data mart proliferation was generally not a chosen architectural strategy. Rather, it was a consequence of not selecting an architectural strategy.
Data mart consolidation is the process of removing data from disparate, non-integrated data marts and placing the data into a centralized, integrated data warehouse. From an enterprise perspective, DMC allows you to meet the needs of a much larger user group than any single data mart ever could. By merging multiple pieces into an integrated whole, you can define a valuable corporate resource that eliminates confusion and redundancy while nurturing innovation through a consistent, reliable source of information - all while saving millions of dollars per year!
For example, last year, a large bank reduced operating costs by 64 percent by moving toward a single data warehouse platform. A manufacturer recently consolidated 20 data marts in less than a year, allowing them to now save $2 million per quarter moving forward!
DMC significantly reduces operational costs and directly impacts the bottom line. As noted, DMC projects often pay for themselves in less than a year! Once payback is achieved, the cost savings continue to accrue and serve as a source of funds to consolidate additional marts. This allows the data warehouse to expand as the number of marts consolidated increases rather than making the full investment up front. Successful DMC projects are cyclical in nature in order to satisfy financial constraints while keeping the complexity manageable.
We can categorize DMC projects into two categories: rehosted or re-architected. In many cases, these categories can also represent two phases of one large project. Rehosting refers to the process of taking database designs and processes off of one platform and placing them on another. This is consolidation in its most raw form. While there are some cost savings achieved with rehosting, the real value will be apparent once the re- architecture is complete. Re-architecture refers to the merging of two or more application-specific databases into a single enterprise database.
While the cost savings opportunities associated with rehosting can more than cover the cost of the project, re-architecting can deliver astronomical value. Imagine being able to ask any question, of any data, at any time. For example, what would be the impact of an ad hoc query that compares your profitability per customer analysis to the customer attrition forecasts? No reconciliation, no merging of spreadsheets, no questions about accuracy.
Figure 1: Savings for Data Mart Consolidation
Assigning ROI metrics to silver- bullet queries requires a fair amount of estimation. In sticking with the introduction of this article, we will focus on tangible metrics. That way, if you can justify DMC on the basis of hard-dollar cost savings, any upside related to integrated information can be considered financial "gravy."
The proven results of DMC are appealing. Eliminate data redundancy and inconsistency. Reduce the number of systems to manage. Lower maintenance costs. But how do you get started? How do you turn a major industry trend into a reality within your organization? The key to a successful DMC initiative is to transform a concept that has been proven elsewhere into a financial model specifically tailored for your business. After all, while it may be impressive that others are saving millions of dollars per quarter, you still need to address the inevitable question of, "What does this mean to me?" While these models can be created at various levels of complexity, let's examine a rather simple formula for forecasting your specific results.
You begin by prioritizing a subset of marts that are candidates for consolidation. This prioritization can be based on a number of factors such as upcoming lease expirations, high levels of data redundancy with other systems, groupings of systems with similar data definitions, or performance degradation due to data and/or usage growth.
Once you've identified a manageable set of marts, you can begin assigning costs to each mart. Begin by estimating the number of full-time equivalents (FTEs) that are associated with the system. This is typically detailed as a percentage of individuals' time spent on the systems being evaluated. For example, if a database administrator (DBA) manages six systems but spends 75 percent of his/her time on the three systems being evaluated, you would allocate 75 percent of the DBA's fully loaded cost to the DMC financial model. Then, follow the same process for ETL programmers, application developers, support personnel, data modelers and any other group involved with supporting the system. Next, locate the non-personnel charges such as real estate allocation, depreciation, maintenance fees and networking costs. You will probably have to contact your finance department rather than IT management for these numbers. As stated earlier, the cost per mart will typically be $1-2 million. The actual total may surprise you, as these costs have historically been spread across disparate budgets and not readily available.
I know what you're thinking. How can you count 75 percent of someone's time as cost savings when they are still being paid as a full-time employee? From a financial perspective, you have eliminated redundant activities and the costs associated with them, not the individuals. Nonetheless, these are tangible cost savings as you now have available resources to apply toward more valuable projects. This is the basis for driving additional business value.
By liberating resources (time and cost) from mundane, redundant tasks, you have created a resource pool to improve business insight, enable faster decisions and gain a competitive edge.
Now you'll need to estimate the total support costs for the future consolidated data warehouse. Assuming that you'll follow the re-architected approach, you can estimate these costs with a high degree of accuracy. You now have a single database to manage, a much smaller number of ETL programs to maintain, and new user requests are handled much faster. After all, new applications can just be "views" of the underlying data warehouse, not physically separate systems. New user requests would no longer require separate ETL processes; the data is ready for use.
The final piece of your DMC financial model is the most difficult. It requires estimating the cost of the hardware, software and personnel (internal or consultants) required to actually perform the consolidation. Keep in mind that the consolidated warehouse will most likely be developed on a platform designed specifically for data warehousing and will therefore require significantly fewer resources than the previous generation of marts created on OLTP platforms.
You may want to enlist the help of someone from your finance department. In essence, you will need to compare the cost of supporting your existing environment over the next three years to the forecasted cost of supporting your new data warehouse over the next three years, including all of the estimated costs to actually perform the consolidation. All you'll need is your corporate tax rate and cost of capital and you're ready to compute the net present value (NPV), internal rate of return (IRR) and payback period for your DMC project.
As with any investment, we must address the fact that there are limited financial risks associated with DMC projects. The good news is we can define the vast majority of risk in two words - political resistance. Despite documented, proven financial benefits, some data mart owners are less than enthusiastic about the idea of consolidating their system into an enterprise data warehouse.
In fact, respondents to a recent Teradata study, fielded by BuzzBack online market research, labeled "organizational politics" as the greatest obstacle to data mart consolidation. Response rates for political barriers dwarfed those of barriers related to data integrity, system incompatibility, technical capabilities and resources.
Data mart consolidation will not strip data mart owners of their ability to manage their content and data access rights. Successful DMC projects have leveraged database views to develop virtual (or logical) data marts, as well as establish user access requirements. This approach to data integration will allow you to achieve economies of scale by implementing a "load once, use many" approach while still providing data mart owners with control procedures.
As with all IT projects in 2003, financial metrics are vital to the success and continued funding of your DMC initiative. The most successful projects have demonstrated value every 60-90 days. While the initial financial justification can be built on a large scale (possibly 10-20 marts), you cannot afford to wait until consolidation of all marts to claim success. By demonstrating value every quarter, you will drive internal momentum and overcome political resistance.
There are too many benefits of data mart consolidation to list in this brief article. In 2003, the majority of organizations will explore the financial benefits purely from a cost savings perspective. However, the true innovators also recognize the revenue/margin uplift potential of an enterprise data warehouse funded by DMC. When the economy inevitably picks up, these organizations will be in a position to leapfrog the competition. Their path to that enviable position will begin with a plan that reduces identifiable redundancies within their existing infrastructure.
Todd Higginson is senior marketing manager of professional service for Teradata, a division of NCR. He is a frequent speaker on the topic of quantifying the return on data warehousing investments and the process of consolidating disparate data mart environments. He can be reached at email@example.com.
|View Full Magazine Issue|
|E-Mail This Article|