Analyzing the economics of IT workloads deployed in Public, Private and Multicloud
“The cloud” has transformed IT and enterprise digitization over the last decade, buoyed by public cloud services that are extensive and continually growing. There are essentially three types of cloud deployments that an enterprise or software-as-a-service (SaaS) provider can employ to run their different workloads. They are:
- Public cloud: AWS, Azure, Google Cloud Platform, etc.
- Private cloud: Privately owned and operated data centers, either on-premises or in a co-location (CoLo) facility.
- Multicloud: The use of any combination of public and/or private clouds. Most observers refer to “hybrid cloud” as a specific type of multicloud that includes private and public clouds.
Over the past ten years, most organizations have migrated at least some of their data and applications to public clouds. As organizations explored the various clouds that best suited their needs, they may have come to favor one deployment model over another. For many years, Juniper Networks has been advocating that organizations employ a multicloud strategy while managing resources as a single, cohesive infrastructure—regardless of whether they reside in a private cloud, public cloud or hybrid cloud. Any debate that may have ever existed over public vs. private cloud has been largely settled: As noted in Flexera’s 2020 State of the Cloud survey, 84% of enterprises use a multicloud approach.
However, at the application by application level, most enterprises and SaaS companies are still struggling to determine the optimal place to run their various workloads. Figuring out the answer to this seemingly simple question is actually quite complex and challenging. The crux of the issue is determining how to optimize IT strategy by keeping expenses low while at the same time maximizing availability, performance and business agility. But inadequate knowledge about different offerings and pricing for cloud models presents a barrier. In addition, larger firms in particular may not even have a clear view of the hundreds of different applications they are currently using to run their business.
We have found that most customers do not use sufficient rigor to analyze the economics of different cloud models before making a decision. Too many companies are making major strategic decisions without adequate analysis and supporting data, impacting the business for many years to come.
The Cloud Workload Decision Model
To help our customers make informed data-driven decisions, Juniper partnered with ACG Research to build a detailed financial model – the “Cloud Workload Decision Model”.
The financial model is an Excel TCO calculator that compares the expenses of running different types of applications and workloads on public cloud vs. private cloud infrastructures. Although there are other cloud calculators available today, they lack the thoroughness required to evaluate the true costs of public vs. private clouds. This is the motivation behind our model which provides a detailed comparison of AWS, Azure, private cloud and hybrid cloud expenses for a combination of ten different application use cases. The model calculates outcomes based on a comprehensive set of input variables, parameters and calculations that cover all associated CapEx and OpEx. Additional considerations beyond the numbers certainly do exist, including centralization/control, security, compliance and regulations, but the model forms the foundation for a company to make sound cloud decisions.
To learn more about the financial model and our detailed findings take a look at the white paper. The paper provides an overview of public and private cloud economics along with detailed comparisons of monthly expenses for different use cases and applications running on AWS, Azure, private and hybrid cloud.
Private vs. Public Cloud
We’ve found that by running different workloads in different clouds, our customers can optimize cost, scalability and flexibility. In other words, it is all about the right cloud for the job.
In general, applications with modest or intermittent CPU requirements (i.e., web servers, application development and testing) are far more cost effective on a public cloud. But applications with high data transfer requirements (i.e., network functions virtualization (NFV) and video gaming) can typically be 10x more expensive to run on a public cloud. There is no doubt that these high traffic costs back to centralized clouds are driving many edge cloud buildouts by service providers and other entities. Juniper is laying the foundation for cloud-neutral edge ecosystems by partnering with StackPath. Finally, applications with GPU (graphics processing unit) compute requirements (i.e., machine learning and artificial intelligence) are also usually much more expensive to run on public cloud vs. private cloud.
Our cloud workload decision model contains numerous dials and knobs that we can tune and turn to modify input variables and calculations to drive an analysis that most accurately reflects a particular customer’s environment. In addition, our analysis is based on AWS and Azure list prices—heavy public cloud users likely negotiate long term volume discounts.
The Future is Indeed Multicloud
Our key finding is that multicloud deployments offer enterprises and SaaS providers the optimal way to manage IT. The future is indeed multicloud and now, we have built the model to help you make the granular, application by application decisions. By identifying compute intensive applications and running them on privately owned data centers, organizations can optimize costs and retain more control. On the other hand, applications with modest compute and memory requirements needing only intermittent access to resources are very cost-effective on public cloud deployments.
Depending on the applications and use cases, we have shown that customers can realize cost savings of up to 91% with a multicloud/hybrid cloud solution. For this reason, it is critical that enterprise and web scalers use a unified software stack that includes networking and security, such as Contrail Enterprise Multicloud, to facilitate workload portability between public and private clouds. The biggest advantage of public cloud services is the inherent agility they offer – this must not be compromised by a multicloud strategy.
If you are an enterprise or SaaS provider and have questions about how to start your cloud journey, where to run your different workloads or if you simply want to better understand the economics of public, private and hybrid clouds, contact your Juniper Account team for a consultation. Our model can be customized for your business conditions. No matter where you are in your cloud journey, we are here to help.