Cost savings is one of the key drivers behind cloud solutions. In this blog, let us explore an approach to perform Cost Benefit Analysis of cloud computing.
Cost Benefit Analysis
Depending upon the type of cloud (IAAS, PAAS, SAAS) and structure ( Private, Hybrid, Public), the cost-benefit modeling can be complex. It is essential that multiple “what if” scenarios are considered.
The steps conducting cost-benefit analysis are:
- Understand the true cost of the current service
- Understand the true cost of the cloud service
- Compare for different scenarios
Understand the true cost of current service:
The objective of this step is to estimate the “unit cost” of a service – for example, payroll processing, procure-to-pay, messaging service, CRM etc. A service will be provided by people and supported by applications, computing infrastructure. In addition, we need to consider the space, electricity and support cost associated with the service.
The following figure shows the components of a service.
Finding the “true cost” of a service can be effort intensive. Many organizations may not have operational cost of buildings and may not track the occupancy by departments. The application, infrastructure cost estimation assumes that the organization captures the incidents/service requests at a Configuration Item (CI) level. However, it is important to get the “true cost” of the service estimated before comparing the benefits of a cloud service.
Understand the true cost of cloud service:
The cost components are:
- Cost of transitioning to cloud environment
- Operating costs in cloud environment
- Transitioning out of cloud environment
Cost of transitioning to cloud environment:
When there are legacy applications and data careful planning is required. The problem is not different to outsourcing, but may not have the same level of vendor support. For example, in a typical outsourcing contract the external vendor will perform the “due diligence” of assessing the infrastructure and migration. Even if the vendor may provide this service for a cost, the organization need not own the migration task.
In a cloud computing arrangement (assuming public cloud), the cloud vendor may not take the responsibility of data migration and application transfer. As a cloud vendors specialize in leveraging the “economy of scale”, they may not provide organization specific services. The organization may have to own the data and application migration cost.
The costs associated in this steps are:
- Contract fees (if any) to terminate the existing legacy applications
- Personnel cost for due-diligence planning
- Personnel cost associated with data migration
Typically when a large organization takes up an outsourcing exercise, they will discover many bitter truths. For example, there can be lot of data duplication without any integrity. Each business units may have their “pet version” of a database. For moving to an integrated cloud, there can be additional costs to prepare the existing information infrastructure.
Operating costs of cloud:
The operating cost of cloud depends upon the licensing model. Some of the possible licensing models in the SAAS cloud environment are:
- Named licensing
- Concurrent licensing
- Transaction based licensing
- Capped licensing
- Unlimited licensing
We need to understand the demand, the licensing model and determine the cost. Please see my other blog on demand management to highlights the fundamental difference in the “owned” infrastructure costing and “cloud infrastructure” costing.
Let us workout the operational cost of a transaction based application. Let us assume there are 2000 users that will be using this application. At any given time there will 500 users using the application. The application is used to complete a transaction (say logging a service request). There are 3000 to 4000 transactions per month (36,000 to 48,000 transactions/year respectively)
The available costing options are:
Named user ( per seat license) at $800/seat/annum
Concurrent licensing cost of $400,000 which allows 500 users to access the application at any time.
Transaction based cost of $10/transaction. If we assume our transaction is a Service Request, every time a person logs a Service Request $10 will be charged.
Assuming a vendor gives a Capped transaction licensing. The organization can log upto 4000 transactions a month for a flat annual fee of $35000. If they go over the Cap, they will charged at a rate of $20 a transaction.
The following table shows the different costing options and estimate.
|Organisation data volume assumptions|
|Number of users||
|Number of average transactions per year||
|Number of Max transactions per year||
|Named license cost per user||
|Concurrent license for 500 users||
|Capped transactions cost (4000 tr/month)||
|Additional transaction cost over Cap||
|Costing Estimate for different options|
|Transaction license (4000 tr/month)||
|Transaction license (5000 tr/month)||
|Capped transaction for 4000 transactions||
|Capped transactions for 5000 transactions||
At a first glance you may conclude that Capped transaction is the cheapest option. You may be correct, if you can limit your transactions to 4000 transactions a month. However, if the transactions go to 5000 transactions/month, the capped option is not the cheapest one.
The problem with this approach is that, the IT department may direct the users not use the application to save operational cost. Unfortunately this is not the right approach as benefits associated with the application will not be utilized. (for example, instead of raising a service request, you may send an email. The efficiency of online processing will be lost if we use a manual workaround to save cost).
The same argument holds good if you choose “pure transaction” based licensing. For 5000 transactions a month, it is cheaper than exceeding the Capped option. ($480,000 versus $590,000)
If the application is not business critical, the optimum option is to choose concurrent licensing for an annual fee of $400,000.
In summary, the usage based model alone may not be cheaper. You need to carefully consider the best option for your specific organisational needs.
Transitioning out of cloud computing
Like any outsourcing contract, the transitioning out of cloud need to be carefully planned. It may be complex than a typical outsourcing deal, as the cloud provider will aim for a generic solution. For example, if your data is stored in the cloud and the format has changed, it may not be easy to retrieve huge volumes of data. As the standards are just emerging, there is no agreed protocol to inter-operate between cloud providers.
Compare the cost of in-house hosting versus cloud
Now we know the “true cost” of hosting an application in house and in different cloud models. The business case should outline the cost benefits, other benefits (like flexibility, agility) and analyze the risks.
I will blog about Risk analysis at later stage.