Many enterprises that once looked to the cloud as a way to curb data center expenditures are now beginning to wonder why they’re not saving as much money as they expected.
“There are no guarantees that cloud computing in any form will save you money,” observes David Linthicum, chief cloud strategy officer at professional services firm Deloitte. Bringing cloud cost goals back on track is relatively simple, however. “It’s a matter of advanced planning and setting realistic expectations,” he notes.
Is your organization wasting money by not intelligently managing its cloud commitments? The following seven management tips can help you bring costs under control and closer to the goals you originally expected.
1. Plan and analyze
Cloud adoption, by itself, isn’t an automatic way to save money. Thorough planning, followed by a detailed cost-benefit analysis, is essential. “Enterprises need to approach cloud computing as a tool that has the potential to save a great deal of money,” Linthicum advises. “However, savings won’t materialize unless you come to the public cloud with clear objectives and a plan around what will be the cost savings and how you’ll get there.”
Start by mapping out current resources, then build a best-guess estimate of what each service’s requirements will be for an entire year, suggests George Burns III, senior consultant of cloud operations for IT professional services firm SPR. Then take the planning a step further and decide how long specific resources will be needed. “Finally, use this information to purchase reserved or pre-paid resources,” Burns says. “By committing to a resource and spending level, you can save one-third or more on your total resource spend.”
2. Strive to optimize
When migrating legacy workloads and systems to the public cloud, anticipate your provisioning needs and “right-size” your cloud configuration.
“Once in the cloud, halt the common legacy data center practice of overprovisioning when deploying new resources,” urges Albert Abello, cloud services manager at digital transformation services consultant DMI. “Rely on autoscaling/dimming, serverless offerings and the elastic nature of the cloud to dynamically and quickly respond to fluctuations in demand.”
Abello also advises spinning up test and demo environments only as needed and tearing them down immediately after they are no longer required.
Another wasteful cloud practice is giving teams free rein to spin up new instances. Often this results in instances running long after their need has ended.
“This can be great for organizations with smaller, diligent and fast-paced teams that need to respond to changing circumstances quickly,” explains Robert Sanders, director of big data engineering for IT consulting firm Clairvoyant. Yet teams often forget to perform capacity planning before spinning up or resizing an instance. “This results in a lot of surprise when customers see the bill,” he notes.
3. Create a cost-conscious culture
Moving to the cloud is as much a cultural shift as it is a technology change, observes Vuong Nguyen, managing consultant at IT consulting firm DMW Group. He believes that it’s important to encourage a cost-conscious cloud culture that’s focused on saving money and boosting performance.
“Cost optimization is everyone’s responsibility,” Nguyen states. Delivery teams, for example, should be responsible not just for providing code, but also for managing its costs, both fixed and variable. “They are in the best place to take on this responsibility, as they can make the best trade-offs and optimization decisions related to costs,” he advises.
Both cost transparency and clean financial data are required to provide accurate visibility into spend and cost savings opportunities, says Carolyn Bellisio, technology CFO of Liberty Mutual Insurance. “The public cloud providers supply massive amounts of valuable billing data,” she notes. “It’s important for organizations to establish a strategy to enhance this data with other metadata, including tags, in order to effectively and efficiently use this information.”
Finally, organizations need to measure their cloud spend against business metrics, such as total revenue, subscribers and orders completed. This is important, because it allows you to add context to the spend and change the conversation from cloud cost to cloud ROI, Nguyen says.
4. Go cloud native
Cloud native — an approach to building and running applications that exploits the advantages of the cloud delivery model — is usually the best way to move existing workloads to the cloud. But some organizations opt for a “lift and shift” strategy — migrating workloads from on premises to the cloud with little or no modification. While there are benefits to this approach, such as faster roll-outs, there are also some costly risks, ranging from latency and performance issues to total migration failures.
“These organizations usually end up spending enormous sums with [cloud providers], as … workloads are not optimized for cloud environments,” says Dave Christian, cloud architect for digital business solutions provider Anexinet.
Although sometimes a viable option, lift and shift should be used only after its impact is fully known and understood, cautions Brendan Caulfield, co-founder of cloud consulting firm ServerCentral Turing Group. He notes that nearly all IT leaders have, at one point or another, experienced times when best intentions failed to materialize into real-world accomplishments.
“If you lift and shift your workload into the public cloud with the intention of revisiting those workloads to right-size them for efficiency and cost control, it’s quite likely that once those workloads are humming along in the cloud you will move onto the next task on your list and forget to get back to optimizing,” he says.
5. Get organized
Cloud costs can be minimized by building and maintaining a solid governance regimen. “Partition cloud consumption into business unit–specific subscriptions,” Abello advises. Require naming and tagging standards across all subscriptions. “Enforce those standards, and resource/region restrictions, with automatically validated policies,” he adds. “Utilize built-in auditing, budget constraint and reporting capabilities to ensure compliance.”
Abello says it’s also wise to require teams to properly designate the owner and project relationship of each instance. “This can ensure that when you see a higher than usual bill, you can trace back which instances and, consequently, which team had contributed to the larger bill and identify if [the cost] is really needed,” Sanders explains.
6. Consider a modular approach
By adopting a modular approach — breaking a monolithic cloud venture into micro-services, or even nano-services — cloud compute time can be focused onto workloads that are most necessary.
“Costs can be minimized by enabling … supporting services to only be invoked when they are needed,” says Rajesh Jethwa, delivery director of Digiterre, a software and data engineering consulting firm. “This approach also supports better management of spikey workloads when there is peak demand for a particular service.”
7. Exploit available savings opportunities
An important key to financial success in the public cloud is embracing a pay-as-you-go model while taking advantage of available savings opportunities. “If you manage your public cloud infrastructure like an on-premises data center, you’ll end up spending more money without a doubt,” Bellisio warns.
Bellisio suggests leveraging the different pricing options available from the public cloud providers to reduce costs on resources that need to be continuously up and running. “These pre-purchase reservations allow for substantial savings over on-demand pricing and are a must for any cloud cost optimization program,” she notes.
Enterprises should also take advantage of prepacked and managed resources. “While some managed resources may cost more initially, consider your time savings in being able to dedicate your engineering and operations resources to other tasks and goals,” Burns says. “For example, many database platforms are available as fully managed services where redundancy, upgrades and backups are all performed seamlessly and without user intervention. Removing the management of that database layer from your team can allow that time to be reallocated, increasing productivity thereby giving you a better ROI.”
Colin Toal, CTO of Chisel AI, an AI technology provider serving commercial insurance firms, suggests using cold storage services, such as Amazon Glacier or Google Coldline, to pay less per gigabyte for the long-term storage of non-critical data. “Using the right backup and storage tier based on the speed and availability of the data is important to optimize cost savings,” he stresses.