How to Spend Money Judiciously on Cloud

Cloud is an integral part of every organization’s digital transformation roadmap. The term “zero CAPEX” has often been associated with public cloud platforms. The true benefits of “economies of scale” are passed on to the customers; as a result of which, the costs of cloud services have been continuously decreasing and the number of services have been on the rise. According to Gartner, the public cloud market is estimated to reach $383 billion in the year 2020.

Cloud Computing is financially attractive for a number of reasons. Since the user of public cloud uses space in a third-party vendor’s data center, they save on real estate, equipment, power, backup and manpower. The utility-bill pricing model is another reason for popularity. However, a frequent user can experience a different issue of “Cost Explosion in Cloud”, when they end up buying items they really do not need. The queries that intrigue most businesses are:

  • What does my bill look like?
  • Where is my cost distributed?
  • How can I bring down my bill the next time I shop?

It is a common trend in business that only huge bills demand everyone’s attention whereas the average bills are overlooked. Hence there arises a need to beget a bird’s eye view of what one is spending on. This covers the ‘What’ aspect of cloud spending.

To understand the ‘Where’ aspect of cloud spending, an in-depth analysis of the cloud invoice is required. There are various perspectives for viewing the fine-grained details about cloud expenditure depending on the viewpoint of the stakeholder. For instance, certain futuristic reports are important from the viewpoint of a CEO who wants to make sure that the Return on Investment (ROI) is commensurate with the projection.

Knowing where the cloud costs are distributed is just the beginning of governing the cloud. It is the ‘How’ aspect of cloud spending that delves into the remedial measures.

Cost Control Levers

The first lever of architecting the infrastructure and applications in cloud is an iterative process, wherein a set of paradigms are applied to the design that would make it flexible and conducive to expansion. A mindset of planning the architecture ahead of time to avoid overprovisioning has to be followed by employing auto-scaling and choosing loosely coupled architectures. One should check on newer and cheaper cloud-native services during the architecture phase itself, to bring down cloud expenses.

The second lever of determining the right cloud vendor and services is also key to cost control. The parameters captured during an in-depth application assessment and infrastructure discovery will help determine which particular target cloud platform is best suited from the cost, compliance and technology point of view. A TCO calculator should be used to compare different vendors. Sometimes using a multi-cloud approach could work out cheaper. One must choose the right SaaS/PaaS offerings by vendors and managed services offering license mobility.

The majority of costs in cloud are incurred by the running instances or servers and their associated data storage. Hence, it becomes important to monitor the cloud inventory periodically. Shutting down machines during weekends and week nights, shutting down services that are unused for defined periods of time and avoiding expensive DR approaches, unless absolutely necessary can minimize usage of resources, resulting in cost savings.

From the cost perspective, the relevance of tagging the services cannot be stressed enough. With this information in place, one can know which application is using a service and where exactly the costs were spent. It also makes it easier to determine which servers to downsize when there is more information handy. Automating the process of tagging resources is part of a good governance practice.

When a user has over-provisioned a deployment, it becomes necessary to detect low utilization and re-adjust the machine sizes. Or one could go with the cloud provider’s size recommendations. Cloud metrics give insight on how the machines and applications are loaded that helps in the resizing activity.  The resize activity itself should be done carefully not to disrupt the running environment.

Another cost control lever is to choose a cost-effective pricing plan. The level of commitment and usage in cloud can determine the pricing plan. For example, the pay-as-you-go schemes are used for short-term and pre-paid schemes are used for long-term use cases. Pricing scheme switch-overs are sometimes cost-savers.

The enterprise should have a good cloud governance framework in place. Governance is a broad term but cost is an important aspect of it. The framework should define the maximum spend capacity for all departments to keep costs under control and give access to only legitimate users to provision services for approved business cases.

To sum up the article, the main concern of an enterprise that has stepped into cloud is to get a grip of its cloud expenses. By taking the time to view, analyze and implement the “what”, “where” and “how” aspects of cloud spending, organizations can achieve highly cost-effective and sustainable cloud adoption.

profile-image

Anita Nair

Guest Author Anita Nair is a Senior Cloud Architect at UST Global. She brings in several years of experience Architecting, Designing, Developing and Testing Embedded Applications in the C/C++ domain. She is passionate about keeping up with the uptrends of the latest technologies in the world of Computing. Her current projects have been on Cloud Computing with AWS, with special focus on Cloud Finance, Cloud Governance, Cloud Security and Cloud DevOps.

Also Read

Stay in the know with our newsletter