Why does on-prem look cheaper? Gartner seems to be able to say “the cloud is cheaper”. If on-premises looks cheaper, then the comparison is often wrong. If you are a very large company and have a VPS rental server like AWS lined up for emergency power generation and flood protection, and you are building a full-fledged server room with mission-critical non-stop servers by installing LINUX on a 1-chip mainframe on AWS or a small 19-inch IBM product mainframe, then you are probably not going to be able to afford on-premises servers. I think it would be a different story if they were to build a full-fledged server room. Translated with DeepL.com (free version)

https://www.sbbit.jp/article/cont1/141101 

In the debate over whether to continue with on-premise or move to the cloud, cost is a key consideration. Many companies have actually moved to the cloud, only to find that the cost was higher than expected. Gartner has concluded that the cloud is cheaper than on-premise environments. Gartner explains six reasons why on-premise seems cheaper and how to effectively utilize the cloud.

Writer: Jun Onaga, freelance writer

What are the reasons why the cloud is cheaper than on-premise?

(Source: Gartner (December 2023))

< Table of Contents

Six reasons why on-premise appears to be cheaper

On-premise may not be “that much cheaper

Why are companies “overspending” on the cloud?

Cloud cost savings won't happen instantly

What I&O leaders “need to do” now

6 Reasons Why On-Premise Seems Cheaper

　According to a Gartner study, the cloud is cheaper than on-premise environments. I will now explain why. Exceptions do indeed exist. But I have to tell you that there are more people who think they are the exception than the real exception,” said Autumn Stanish, Senior Principal Analyst at Gartner.

Gartner

Senior Principal Analyst

Autumn Stanish

The only exception is for special workloads that cannot run outside of an on-premise environment. Unfortunately, many workloads are not that special,” said Stanish.

　So, Stanish asked, “Why does on-premise appear to be less expensive?” Stanish explains that there are six reasons.

　The first is that the comparison is wrong.

　When comparing on-premise environments to the cloud, he is often asked, “Which is cheaper for running virtual machines, on-premise data centers or the cloud? Stanish points out that making such comparisons is wrong in itself.

　Gartner defines the cloud as “a computing style that leverages Internet technologies to deliver scalable and extensible IT-related functions as a service” to customers. As such, Stanish explains, “You can't compare a service to a physical server.

　He also said that there are several options for cloud as a service, which will enable customers to take the first step in moving to the cloud. Specifically, “Function as a Service (FaaS),” “Step Functions,” “Serverless Containers,” “Hosted Kubernetes,” and “Cloud VMs/Instances,” each of which has different operational complexity and costs, he said.

Shifting to the cloud isn't basically just about lifting,” he said. First of all, we should think about cloud migration differently than we have in the past,” Stanish said.

On-premise may not be “that cheap.

　Second, “On-premise costs are underestimated,” said Stanish.

　In fact, it is very difficult to properly analyze the cost of a data center,” said Stanish. Also, factors such as internal politics can come into play, and it is possible to estimate your department's numbers to make them look as good as possible,” he observed.

　But it is also important to track all serial numbers and get the cost details down to the level of confirming the price at the time of purchase. We need to take a fresh look at the infrastructure itself,” Stanish said.

　On the other hand, “Many people don't understand maintenance, and few are willing to go through all the service desk tickets to find out how much they are actually costing. In other words, there are a lot of complexities involved in analyzing infrastructure costs, and frankly, you don't have the time to do it yourself,” he noted.

　Stanish recommends using Gartner's sheet for analyzing the full cost of an on-premise data center. The sheet, he said, will help you understand how to create a data center cost model suitable for comparison with public clouds, and will give you an accurate idea of the cost of your data center.

　And third, it does not take quality into account.

　Stanish explains, “It's hard to get the functionality that cloud providers offer in our own environment, but the cloud has a lot of different features that you can't get on-premise.”

　In fact, many of Stanish's clients do not even have the basic automation that the cloud offers at this time. By moving to the cloud, we can remove that complexity,” Stanish said.

What the cloud can do that on-premise can't

(Source: Gartner (December 2023))

　Mr. Stanish then brought up the story of one of his colleagues when he was a server engineer, when he was trying to add redundancy to a data center. According to him, “A company he worked for once forced us to block a road for four days, lay fiber cable under it, and repave the road again.

Even for colocation providers, building a Tier 3 or Tier 4 data center is not a simple task. Expanding such redundancy is just a click of a button with the cloud. Quality is a very important consideration when running infrastructure,” said Stanish.

Why are companies “overspending” on the cloud?

　Fourth, companies are “overspending” on the public cloud.

　A common mistake we see in the public cloud is overspending a lot,” said Stanish. The main reason for this is that cloud capacity is not optimized. If you operate it in the same way as a data center, you tend to worry about running out of capacity. As a result, they tend to over-provision, but in reality, there is no need to do so,” he explains.

　He also cites three reasons for overspending on the cloud: lack of capacity optimization, lack of commitment management, and cost-related modernization.

Why overspending on cloud computing?

(Source: Gartner (December 2023))

　To remedy overspending, the report recommends “right-sizing and scheduling, and consolidating unit costs through analytics/automation,” “taking advantage of institutional discounts and obtaining private pricing agreements (PPAs),” “auto-scaling, network optimization, storage lifecycle The report also recommends “reviewing the architecture by applying auto-scaling, network optimization, storage lifecycle, and moving to a serverless architecture.

　In addition, the report recommends using Gartner's Cloud Cost Management Framework. They say it can help them achieve specific discounts, private pricing agreements, etc., or look at historical data to understand what they will need in the future to achieve better discounts.

　In addition, to achieve more strategic and long-term goals, it is also important to use the tools needed to optimize the environment provided by the cloud provider.

“Cloud providers don't want you to over-provision and waste resources, they want you to use your data center as efficiently as possible. Utilizing the tools they provide will be the first step in finding the first action item,” Stanish said.

Cloud cost savings will not be immediate

　Fifth, “we are only measuring a specific point in time.

　It is true that the cost savings of the cloud may not be realized immediately,” Stanish explained.

“When you start moving to the cloud, it's going to cost you more. That's because of the infrastructure running in parallel with the cloud and the added cost of migration,” Stanish said.

　And looking back on the first year of operations, “We migrated to the cloud and our infrastructure costs increased significantly. It's easy to say, “It was a terrible choice,” but after three or five years of operation, or any other long-term period of time, the cloud will be used more wisely and costs will gradually be optimized.

As you right-size the workloads you need and stop overconsuming and over-provisioning, and as you stay with your cloud provider longer, volume discounts will apply and your cloud provider will be able to charge you less.” (Stanish)

　The sixth and final point is “ignoring opportunity cost. According to Stanish, this is the most important one.

　By moving to the cloud, the staff in charge will be freed from the exhausting tasks they have become accustomed to, such as “building servers,” “patching servers,” “deploying/updating applications,” “storage management,” and “lifecycle management.

　By moving to the cloud, they can shift to higher value tasks such as automation and DevOps, GitOps and platform engineering. Stanish stresses, “Your own organization can deliver the next level of business value, and your I&O department can gain recognition for these efforts.”

　Gartner predicts that “80% of all enterprise workloads will be deployed on cloud infrastructure or platform services by 2027,” a significant increase from 40% in 2023, “a phenomenon that follows the trends we have historically tracked and We are seeing things accelerating toward a cloud environment.

Many companies and organizations are moving to the cloud. Not everything will be perfect, as there are exceptions, but the future is coming where the majority of workloads are in the cloud. It's better to start thinking about cost-effectiveness now,” Stanish said.

What I&O Leaders “Need to Do” Now

　Finally, Stanish proposed an action plan for I&O leaders.

　According to him, they should immediately implement measures to control cloud costs. Specifically, this means starting with the cloud provider's built-in tools and putting in place budget alerts and other preventive measures.

　Then, they should take actions such as “correctly calculating and accounting for all on-premise spending,” and “modernizing applications to take full advantage of cloud-native technologies,” according to the report.

I would like to see companies understand the potential for cost savings that they have not seen before, such as putting everything that is not a differentiator for them on the cloud,” said Stanish.

This article is a reorganized version of a presentation given at the Gartner IT Infrastructure, Operations & Cloud Strategy Conference held on December 12-13, 2023.

Translated with DeepL.com (free version)