With all the variables involved, managing cloud services pricing is a challenge. Here, we’ll explore ways you can estimate and optimize your cloud spending.
Christina Harker, PhD
Enterprises of all sizes are focusing more and more on reining in cloud costs. In fact, according to a recent cloud budget survey, 64% of senior IT professionals indicated “cost management and containment” was their biggest concern. This makes sense in a world where entire businesses are run in the cloud and cloud services pricing can make or break the bottom line.
However, striking a balance between cost-optimization and performance isn’t always easy. This is particularly true in multi-cloud environments where each vendor’s pricing model and implementation quirks are different.
Here, we’ll take a closer look at cloud services pricing, how cloud management can help, and how Divio helps abstract away the complexity of keeping costs low.
Efficient cloud management – the process of administering all cloud resources and services in an organization – helps enterprises reduce waste and streamline IT operations. Of course, making your cloud management efforts “efficient” is easier said than done.
The complexities of the different public, hybrid, and private cloud environments across different enterprises mean there’s no one-size-fits-all approach that will work for everyone. Even if we hone in on a single public cloud provider like AWS or Azure, the topic of cost-management gets so complex that there are multiple books on the topic.
However, with industry analysts estimating that roughly 30% of cloud spend is wasted, getting cloud management right can significantly impact an enterprise. There’s no single reason for this waste, but some of the most common drivers of cloud waste identified by Gartner include:
Oversizing - It can be difficult to properly size and scale workloads, and engineers tend to err on the side of overprovisioning.
Running workloads when they’re not needed - Many cloud resources are left running – and accruing charges – when they’re no longer needed. This is common for development and test workloads as well as production services that aren’t needed 24/7. Easy fixes exist for this, such as using instance schedulers, lambda functions, and CloudWatch metrics.
Poor design and implementation decisions - Many enterprises don’t have experts for each cloud platform they use on the payroll. As a result, the solutions that get implemented are often more expensive than they need to be.
In addition to reducing costs, cloud management increases the predictability of cloud service costs for better forecasting, improves performance by enabling right-sizing of instances, and leads to better visibility across cloud infrastructure.
A big driver of the challenges associated with managing the cost of cloud services is quantifying everything an enterprise is billed for. The major components of these costs include:
Compute - Cloud compute costs consist of the virtual machines, containers, and serverless functions that run workloads. AWS EC2 instances are the textbook example of cloud compute resources, but anything that provides processing power falls into this category. Exactly how much cloud compute will cost you varies significantly depending on your workload requirements. For example, AWS EC2 Reserved Instances are up to 72% cheaper than comparable EC2 on-demand instances. Spot instances are also significantly reduced in price relative to on-demand instances. Furthermore, the type of hardware you request has a dramatic impact on your cloud compute costs. Using instances equipped with powerful GPUs will be significantly more expensive than using lightweight instances with a single CPU core.
Storage - Cloud storage costs cover all the data an enterprise stores in the cloud. The cost of cloud storage varies depending on the amount of data stored and how it is stored. “Hot” storage – which can be immediately accessed – tends to cost more than “cold” storage that takes seconds to hours to access. Similarly, SSD storage tends to cost more than HDD storage because of the differences in hardware costs. The type of storage (file, object, or block) also influences the cost of cloud storage.
Networking - In addition to egress (outbound) traffic costs, there are a variety of other network charges related to the cost of cloud services. Everything from CDNs to load balancers to public IP addresses can lead to charges on a cloud bill.
Licensing - Licensing costs are often overlooked when computing the cost of cloud services, but they can have a significant impact. For example, licenses for enterprise staples like SAP, Oracle, and Windows products are often a big part of cloud bills.
Backups - Cloud backups are a game-changer from a resilience and disaster recovery perspective. From a cost perspective, they’re a combination of storage, network, and licensing costs. At the very least, enterprises will need to pay for backup storage. When restoring a backup on a new platform, network egress fees also come into play. For enterprises with terabytes of data on their servers, those fees can rack up fast. Furthermore, many backup and recovery services come with license costs of their own.
In addition to these cloud cost drivers, enterprises should consider the less obvious operational costs of the cloud. Most cloud solutions aren’t completely turnkey, and businesses need to account for the associated IT ops costs. While these will usually be lower than running comparable infrastructure without a cloud platform, neglecting to quantify them leaves an organization without a full view of the cost of cloud services.
To create reliable forecasts and make well-informed business decisions, enterprises need to reasonably estimate the cost of cloud services. For the major public cloud providers, online calculators – like those from AWS, Azure, and GCP – are a great first step in the process. In fact, for basic use cases, they may be all you need.
However, for many enterprises, architecture complexity, security, operational costs, and usage variability often mean calculators alone aren’t enough for reasonably predictable billing and estimation. Further, calculating the cost of cooling, power, and ongoing maintenance for enterprises that host private cloud infrastructure on-premises creates additional estimation complexity.
Often, this means estimating the cloud costs is a mix of calculators, license and subscription costs, and internal estimates making their way into a spreadsheet. Done right, creating these estimates can provide enterprises with a reasonable forecast for cloud services, but it takes some discipline and effort.
Want to take the complexity out of cloud cost management? Explore Divio’s simple pricing model for public, private, and multi cloud environments.
If you’re just getting started with estimating the cloud costs, David Linthicum’s CloudOps formula is a useful baseline to sanity-check your estimates. That formula is:
CloudOps annual costs = ((NW*CW)*COM)+((NW*CW)*SR)+((NW*CW)*MR)
NW = Number of workloads
CW = A complexity multiplier from 1.01-2.0
SR = Security requirements rated on a 100-500 scale
MR = Monitoring requirements rated on a 100-500 scale
COM = A “cloudops multiplier” based on resources including cost of services and human costs on a 1,000-10,000 scale.
Estimates based on this formula will not be precise, but they are a good reference for the full cost of cloud services.
Cloud management and keeping the cost of cloud services low are complex tasks. At Divio, we take the complexity out of the process and make it easier for you to build web apps affordably and efficiently.
How does Divio help businesses manage the cost of cloud services? By:
Centralizing multi cloud management within a single holistic platform
Optimizing infrastructure so you don’t pay for unneeded cloud resources
Creating custom solutions that eliminate waste and streamline performance and scalability
Leveraging cost-saving techniques like using affordable EC2 Reserved Instances