It is tempting to think that cloud adoption is the process of migrating from one data centre to another, however, data centre functionalities are only a subset of cloud services....
Pitfalls to Avoid in Cloud Adoption: Spiralling Out Cloud Costs
Welcome to part five of our five part series, Major Pitfalls to Avoid During Cloud Adoption.
We have offered a five part series on the major pitfalls to avoid during your cloud adoption journey, and our suggested solution to each pitfall. Today we’ll cover one of the most hard hitting risks - spiraling cloud costs. It is crucial to understand how cost modelling works when transitioning to cloud, where you may benefit from efficiencies, and how to leverage and manage organizational budgets.
Pitfall 5: Spiraling Out Cloud Cost
One of the major differences between on-premises depreciable hardware and cloud platform is how costs are charged. Most of the on-premises are upfront costs that can be predictably depreciated through their lifetime, while cloud services are primarily based on the pay-as-you-go model and, therefore, can fluctuate significantly based on usage. Regardless of the amount of care in estimating cloud costs, it is rare for cloud costs to align well with projections without tremendous effort.
Solution to Pitfall 5: Allocate Costs Based on Workload and Ownership
Cloud services on a pay-as-you-go model is one challenge, but an even more significant challenge is that each service is charged differently. As each workflow tends to be composed of different cloud services, it takes too much effort to minutely account for different cost parameters and how they roll up to the final bill.
This explains why any cost measurement will be based on allocation. On the cloud, allocation is indicated by metadata in the form of tags. Most cloud adopters are aware of tags, but the real work is in defining how the tags should be tailored for the organization.
At OpsGuru, we argue that the most important tags are the ones that align with budget allocations based on lines of business and P&L owners. It automatically integrates well with existing account practices. Because most resources resource tens of tags, engineering teams then start to include a multitude of tags. The effort is applaudable but difficult to maintain. We, therefore, recommend lower-level tags to be based on workloads – the fewer the tags, the better.
One other challenge about tags will be shared resources or consumption that is difficult to account for. Admittedly for such consumption, approximation is the best way to go by. A dual approach of initial IT budget plus a gratuity-based model (e.g., workload load costs $100 per month, a gratuity of 10% is automatically charged back to the workload owner to cover shared services cost) will avoid surprises.
Cloud costs can be complex for newcomers. This is where collaboration with your company’s finance and accounting team, who are the experts in allocation and budgeting, will be most handy. It is yet another indicator that cloud adoption is not only a technology exercise but an opportunity for cross-team collaboration.