Demand management is to understand the business demand for IT (in terms of bandwidth, memory, CPU capacity, support personnel etc) based on the past business activity pattern combined with the future business growth estimate.
For example, there will be an increase in the transaction volume for a financial services company during the end of financial calendar year. An academic web site needs to cater for increase in transaction volumes, during the beginning of academic year.
Think differently to manage demand in cloud environment!
Why does Apple fall?
Is demand management same for all type of service provisions?
Let us consider Infrastructure As A Service (in a Public Cloud) that hosts a customer facing web application.
In this model, the organization pays only for the resources they consume. The role of Demand management will be limited to estimating the cash outflow. For example, we may look at the monthly transaction volume of last year to estimate the cloud hosting budget for next year.
The technical challenge of managing the peak demand is conveniently outsourced to the cloud provider. As a public cloud, it is assumed that the provider will have adequate capacity to instantaneously increase the capacity.
Instead of worrying about managing the peak demand, we need to focus on reducing the steady (average) demand. In the “non cloud” model, managing the average demand is not an issue as the internal capacity is designed with peak capacity in mind. In the cloud they get charged for the usage. So, if you want to reduce your operational cost, reduce the average demand over time. This may or may not involve reducing the peak demand.
Usage based charging can affect the demand pattern:
Consider a Service Desk Application available to the organization as a “SAAS”. The charging is based on the actual transactions. Before subscribing to this service, the enterprise would have analyzed their historical demand pattern to estimate the cost.
Let us assume, the users love this new application available to them through cloud. The users start logging their incidents and service requests. As a result of this, the Operational cost goes through the roof. If the organization has considered a cloud service only as a cost saving alternative, certainly the initiative has failed. However, if they consider “value” of the cloud service, then it would have been a success.
Do not use only the past demand pattern alone to estimate the cost of cloud service.
If the subscription costs go higher, analyze the reasons behind the increase in usage. Do not blindly try to cut the cost.
The advantage of cloud computing is its ability to manage the changes in demand pattern. If you are business demand does not vary too much and you can predict the demand, probably cloud computing may not required in your business circumstances.