The cloud offers significant cost savings and other benefits. But operating a business in the cloud differs from running a business using on-premises systems. Many enterprises get carried away and overspend on the cloud, negating the advantages.

The cloud spares the enterprise from investing huge capital upfront. But the cloud attracts higher operating expenditures. Service providers bill for storage, network traffic, virtual machine instances, and software licenses. Left unchecked, these amounts add up and become prohibitive. 

A recent HashiCorp-Forrester survey estimates that 94% of enterprises overspend on cloud resources. 1,000+ IT decision-makers across Asia-Pacific, the Middle East, Europe, and North America participated in the survey. Most of them opined their enterprises had avoidable cloud expenses.

Here are why enterprises spend more than they need on the cloud and how to avoid such pitfalls.

1. Over-provisioned resources

Over-provisioning and accumulating underused resources is the number one wasteful cloud expenditure.

59% of the respondents in the HashiCorp–Forrester survey overprovision cloud resources. 66% of the respondents list underutilised resources as the top reasons their company overspends on the cloud. 

The reasons why enterprises end up with overprovisioning are many.

Many enterprises provision for CPUs and memory but never utilise it fully. Scalable cloud services make over-provisioning redundant in theory. But most IT teams carry over the habits accumulated during their on-premises days and overprovision clusters to be on the safe side. They do not estimate the resources needed. Instead, they provide their applications with as many infrastructure resources as possible. 

Many It teams subscribe for capacity but do not discontinue it after use. Orphan services, left over after the resources using it ends, and gluttons or oversized VMs inflate the cloud bill by about 20% to 40%, on average.

Many developers do not have business management exposure. They remain project-focused and provision resources without proper cost appreciation. They may also provision different instances across multiple cloud service providers. The enterprise loses out on volume discounts.

The costs add up as the number of applications, and zombie servers rise. The best way to avoid accumulating unneeded resources is to plan and have good control over provisioning. 

Entrust the DevOps team to make a cost-benefit analysis of resources. Right size provisioning after reconciling development and business considerations. 

• Use resources such as Kubernetes clusters that scale infrastructure resources as required. 
• Deploy newer tools that deploy ML algorithms to synchronise requested and provisioned CPUs over time.
• Opt for containers when the reduced isolation is not an issue. One of the most expensive equipment to lease in the cloud is RAM. Containers need only about one-third of the RAM to run the same software as a VM. 

2. Not managing cloud resources proactively 

The cloud requires strong data governance and controls. Unless enterprises manage their consumption, the costs get out of control.

Identifying underutilised cloud assets is not easy. Bills from cloud providers contain hundreds of line items and are, in any case, reactive. Waiting for the bill will attract steep charges for VMs and other services that were idle for 30 days. Managing multiple clouds with multiple accounts becomes harder.

As effective solutions to cut costs:

• Pre-purchase base capacity at a discount, especially if the expected usage of the cloud exceeds one year. Most vendors have complex pre-purchase plans. Getting bargain prices often depend on making active comparisons.
• Opt for reserved instances for regular or predictable usage. Reserved instances offer up to 75% discounts compared to on-demand instance pricing. 
• Take up spot and low-priority instances, which are often available at discounts of 50% to 90% compared to on-demand pricing. Amazon EC2 spot, for example, allows bidding on spare Amazon EC2 computing capacity. Google likewise offers its excess compute capacity as “preemptible instances.” 

3. Not going beyond file transfers

For many enterprises, the net effect of cloud migration is migration from on-premises to a remote server. The cloud is, however, more than shifting on-premises resources to the cloud and accessing it through the Internet. Cloud cost savings depend on the system’s functional evaluation and re-architecture.  

The cloud centralises information and enables anywhere, anytime access. Success requires transparency and free sharing of information. Without such changes, employees resist cloud initiatives. The costly investments in the new technology remain underutilised. For instance, lack of visibility makes employees use old systems, duplicating the costs.

• Assess the existing data footprint and the comparative cloud need. Re-architecture the business process before moving major business systems to the cloud 
• Opt for serverless cloud computing. Functions-as-a-Service (FaaS) reduces the cost and effort of migrating loads to the cloud. 
• Promote agility. Making the enterprise agile is the prerequisite to profitable cloud adoption. Without enterprise systems and the workforce being flexible, they will not be in a position to take advantage of the cloud offerings.
• Develop a solid CloudOps and FinOps culture and practices. Without such a mindset, enterprises end up making errors and duplicating efforts,  adding to the costs.

4. Lack of skills

Many enterprises that subscribe to cloud services do not have the requisite skills to leverage the full features of the cloud. They end up paying for resources they do not use. In the Hashi-Corp-Forrester survey, 49% of respondents cited a lack of skills as a key reason for overspending on the cloud. Almost 86% of respondents depend on cloud operations and strategy teams to perform essential tasks. They rely on such teams to standardise services, centralise security, and implement best practices.

The crippling skill shortage harms enterprises. Skill development lags technological developments, and the gap is widening.

• Sync cloud investments with skill availability. 
• Have a plan to train the workforce or make new hires to operate the latest resources. Many IT users are unaware of the ability of new services such as Kubernetes to execute active and near real-time rightsizing.
• Promote a cloud-first mindset among the workforce.

5. Not having a plan to control unexpected costs 

Many enterprises jump into the cloud bandwagon blindly. They underestimate the depth of the migration exercise. 

For start-ups, building a technology stack in the cloud is easy. It is much harder for established companies running on-premises systems to shift to the cloud. Unexpected hurdles and headaches add to the cost. 

Migrating to the cloud requires re-conceptualising the architecture and redoing the tools. Many enterprises underestimate the cost of data transfers, retooling, and associated costs. Most enterprises will bring in a new ecosystem of vendors and products, adding to indirect costs.

The most common underestimation is the cost of transactional data storage. Moving an ERP system to the cloud generates 10x data. A new cost transaction generates every time a piece of inventory moves. Customer information, cases, opportunities, leads, and other data soon add up.

Another considerable unexpected cost is revising licensing agreements. Many enterprises negotiate poorly or do not have clarity on their use cases. They pay more than necessary for licences.

The cloud also increases governance costs, which many enterprises underestimate.

Solutions to pre-empt such unexpected costs include:

• Refrain from moving everything to the cloud. Moving everything to the cloud may offer benefits but often does not make financial sense. 
• Prepare for cloud migrations before starting the actual transition. Frame realistic expectations of consumption and usage to prevent sticker shock.

Migrating to the cloud is a double-edged sword. It unlocks several possibilities, but it also mandates new ways of work. Enterprises that manage their cloud resources reap cost savings and become competitive.