Reengineering the IT Supply Chain

IT has spearheaded a major evolution in supply chain efficiency, so it’s paradoxical that provisioning IT resources, which is, after all, just an internal supply chain, is one of the most inefficient processes at most companies.

Many, if not most of the IT supply chain’s inefficiencies and cost overruns stem from a lack of transparency. Between IT providers and business users lies an opaque layer that separates resource needs from provisioning reality. On one side, IT maintains that business users don’t understand the cost of maintaining an IT infrastructure. On the other side, business users say IT doesn’t understand the often unpredictable ebb and flow of the business cycle and its implications for IT.

Neither side offers the other sufficient information to paint an accurate picture of the demand for, and supply of, IT services. This lack of information leads to dysfunctional behaviors that ultimately drive up IT costs. For example, users over-request resources since IT can rarely provision resources on demand; hence, server utilization levels are low at most companies.

So what can companies do to correct this problem?

Ultimately, companies can reengineer their IT supply chain to create an effective exchange of information between all stakeholders. In a traditional sense, Dell Computer’s external supply chain is perhaps the best example of what internal IT reengineering can accomplish.

Dell and others recognized that a supply chain is not just the physical movement of a product from a supplier to a customer. Information is also flowing in the opposite direction. In essence, information around capacity planning and forecasting is a substitute for the physical product. It is only in cases where information is lacking that firms have to hold buffer stocks (additional stockpiles of inventory to be used when forecasting is incorrect). Over time, Dell has spent considerable effort developing models to limit inventory holding costs or to minimize the risk of stock outs.

The IT supply Chain

Provisioning and using IT infrastructure presents dilemmas similar to external supply chains logistics. When hardware (storage and application servers especially) were perceived as inexpensive, firms may not have been overly concerned with the added costs of holding reserve capacity to act as a buffer stock. The demands of applications like CRM mean that users could expect volatility in their storage utilization or processing resources. At certain times they will need additional resources, so firms may need to purchase resources to meet peak demands, knowing that those resources may sit idle outside peak periods.

The processes used to provision resources are equally at fault. Data storage provisioning is especially problematic. For example, an application owner might predict that 20 terabytes are sufficient to support a marketing application. Knowing in advance the difficulties in provisioning larger amounts of storage on short notice, they instead demand 35 terabytes. IT has come to expect some degree of padding and so the ultimately provision 30 terabytes. The net result, however, is low storage utilization and higher than necessary costs.

The underlying problem in this scenario is that IT is unresponsive to sudden changes in resource demands, and so just as in a physical supply chain, users learn to stock up in anticipation of a rainy day. IT cannot provision network resources, data storage or processor capacity at a moment’s notice and so padding is a natural response.

Can IT Be Reengineered?

While companies can reengineer their physical supply chains in several ways, the IT supply chain is a totally different matter.

One way to reengineer the IT supply chain is to ensure an effective exchange of information between all points. Information is freely exchanged regarding cost parameters, service expectations, and service performance. If users believe their requests will be handled in a timely manner, they may be more inclined to issue realistic resource requests. Similarly, IT may be less inclined to assume that all requests are exaggerated, and so they will not waste as much time and effort on managing excess capacity.