Cloud Cost Management

With cloud gaining traction across most enterprises, Cost management is critical aspect to drive adoption of cloud without adversely effecting IT budget. Cost is important pillar considered as part of architecture review process by cloud vendors themselves e.g. AWS well architected framework has cost management as one of the pillar. In many organizations, movement to cloud is driven by cost optimization or reduction that it provides so it needs to be managed well to drive RoI that happens quicker than unmanaged usage of cloud services.

Managing the cloud cost is very different from typical approach taken at on-premise infrastructure management. However at the same time, there is a need to bring in some elements of on-premise infrastructure planning and utilization practices to public cloud environment as well e.g. identifying the services that will be leveraged from cloud and how cost will be tagged for each of these.

There are primarily two types of cloud adoption models prevalent in most Enterprises

• IaaS - Suitable for applications that are currently hosted on-premise and likely moving to cloud . Many enterprises typically use compute services while moving these applications to public cloud. Primary UOM here will be VM for costing.
• PaaS - Suitable for green-field application for sure. It is important to look at on-premise applications and identify patterns that can benefit from adoption of cloud native services. Adopting these services can help you move into transformative adoption model for cloud and also reduce your operating efforts in managing these applications. There needs to be constant effort in refactoring applications on cloud to adopt more and more native services to reduce dependency on purely using VMs or compute services at infra layer for public cloud

You need to look at cost management as both top-down initiative with budgetary provisions as well as establish bottom-up accountability in using cloud services in responsible fashion. Cost management principles needs to be considered at inception phase of the project than coming up details while application has already gone live. Making changes can be expensive at later stage and will likely face inertia.

If we typically look at comparing cost of Infra model vs serverless/cloud native service adoption, there can be significant difference between the two. I have seen difference of around 50%-60% where Infra model is expensive but this is primarily driven by application architecture and use case.

From environment perspective, you need not focus on managing the cost for your production environment alone as other environment cost are important depending upon the development team distribution and size. Right sizing all environments is must for driving efficiencies.

Lastly work with your cloud service provider to identify right kind of pricing models suitable for your applications. You might find on-demand instances are expensive compared to reserved instances or spot instances but in many cases on-demand instances are your best best. If your application architecture permits, you can benefit from usage of spot instances with right right strategy in place.