One of the hardest tasks for IT leadership is considering the cost of IT services against their theoretical return. Because of that, there’s one question we see from enterprises: What is the return on investment I get from transitioning my infrastructure over to the cloud? It’s a powerful question and not just one that the CIO or CTO is asking. Everyone from sysadmins to help-desk teams need to know what the impact will be. So how does that translate down the IT chain?
CIOs ask – Can we reduce our capital expenditure by moving parts of our infrastructure to the cloud?
IT asks – Will we be able to improve service delivery for end-users?
Admins ask – Will it be easy to manage physical servers, server hosts for VMs, and cloud infrastructure? Can we increase our operational efficiency?
Developers ask – Is this going to happen, or do we have to go back to secretly using cloud resources within our teams?
The answer to all of those questions is yes. In exchange for setup, user education, and integrating infrastructure into your existing security protocols – a mandate for cloud is universally beneficial. I came across this read that illuminated the costs of running a SaaS product on the cloud. It highlighted the other big question we hear from enterprises approaching cloud: how are these new operating costs divided? Typically in three ways. Platforms (AWS) to software (individual tools like ServiceNow or PagerDuty) to pieces of infrastructure (like Compute Engine on GCP or individual services on AWS).
If we apply the above formula for ROI, our gains from the investment of cloud would be saving money and accelerating productivity. On the flip side, our costs would be time of integration and education on new tools.
Imagine that you could easily scale products to meet demand without any reduction in service or reliability. As usage slows overnight, unnecessary VMs could be shut down automatically. During on-going periods of increased usage (like tax season), you can provision dedicated VMs that scale automatically. Even frustrating and slow processes like requesting servers can move along faster. And because our infrastructure is operated with no overhead, we’re paying less on upkeep. Not only do we remove ourselves from managing physical infrastructure, but the simplicity of operating costs divided into platforms, software, and infrastructure helps everyone.
Switching over to cloud providers isn’t a panacea. If your tools and workflows are scattered, it’s up to the company to unite them using a cloud management platform or rely on the tools inside one provider like AWS. But a warning for teams that build out their own solution:
- “Independent teams will work on bits and pieces of an automated workflow for their applications, unbeknownst to them that they are actually in the early phases of building a CMP from scratch. “
At some point the CIO finds out that there are dev teams working on redundant implementations and then requests consolidation – to which they’ll have to build an API, UI, and documentation for. Plus they’ll have to maintain ownership and update it to match new AWS APIs and functionality. As a developer I understand the desire to build out a DevOps toolset just the way you like it – but I also know the struggle of wasting a summer building out a tool that turns out to be a major bottleneck. Hence the beauty of CMPs.
Whatever you decide on, there’s real ROI on cloud infrastructure. Let’s look at cloud from the gains and costs where your bottom line can be affected. Using this information, you can calculate your own ROI.
Quantitative Costs of On-Prem
Let’s start at the beginning. What are the units of cost involved in running infrastructure on on-prem? In other words, the things that we pay for as part of capital expenditure.
- Physical servers (non-virtualized)
- Physical server hosts used for virtual servers
- Licenses for server VMs on physical hosts
- Annual maintenance and support for physical servers
- Annual software costs for server machines (hypervisors, SaaS tools)
- Hosting/Physical Space costs (physical building, power, cooling, etc.)
- Ongoing server hardware costs (upgrades, updates, patches, etc.)
- Labor – IT system admin hours, tech support and end-user support
Quantitative Gains Of Cloud
The financial gain of cloud is that computing cost is shifted from a capital expenditure to an operational cost because the cloud provider supplies the underlying infrastructure. To cut back on jargon – operating overhead falls on the cloud providers.
- Save money and efficiency by automatically reclaiming inactive machines after they’re no longer required. So consider what percentage of your VMs are temporary (only used for weeks or months, or during business hours like 8-6)
- Automatically eliminate machines provisioned without justification – skunkworks/over provisioned resources
- Use on-demand resources to eliminate the operational cost and headache of capacity planning.
- Get better performance and up-time from consistent monitoring by the cloud providers or SaaS tools geared for the cloud. (like CloudWatch/CloudTrail on AWS)
- Automatically reusing temporary resources through templating
- Reduce the number of your sysadmins and reduce the annual cost per admin. While you may increase average number of VMs managed per sysadmin, efficiency increases.
- Limit resource consumption by delivering machines at the right size and service level.
Qualitative Costs Of On-Prem
Qualitative costs are the ones you may not be able to directly place on a ledger, but can affect the capital and operational expenditure of a company.
- Machines that exist without proper business justification – either skunkworks or abandoned application fleets
- Machines that are over provisioned – meaning that they have bigger CPUs, memory, or storage than necessary
- IT spending time to fix low-level issues caused by human error
- Slow service delivery to end users – slowing down developers, engineering teams
- Machines that are online and maintained after they are no longer required – leading to wasted resources and labor
- Scalability is complicated because it tends to require planning in advance
Qualitative Gains Of Cloud
- Reduce manual effort due to automation and lower errors caused by manual processing
- Reduce sysadmin work by automating the end-to-end service delivery of servers, creating repeatable processes that automate provisioning and management of VMs.
- Self-Service and improved Service Delivery – which boosts operational efficiency and boosts innovation
- Efficient management of resources boosts productivity. Development + testing is faster, leading to applications deploy faster.
- You can eliminate the use of multiple management tools. As developers change roles and teams over time, the standardization of cloud resources means that new users can onboard to projects faster. Not only do you have the ability to spin up new infrastructure, but you have more visibility on all sides of an application. For example, when you join an ongoing project with no visibility on how the product was configured, having a single place to see where all tools, configuration management systems, SaaS products, microservices, even passwords for logins fit in.
- IT resources can be allocated to support core business functions.
- Cloud can increase collaboration between the enterprise and its customers or reduce response time to customer inquiries. Plus if anything goes wrong, disaster recovery can occur in a more efficient/automated manner.
Costs Of Cloud
You didn’t think this would be a one-sided argument, did you? There are costs involved with using cloud. Here’s some highlights that we thought it would be important to mention.
- Pricing of individual services can get convoluted – We noticed this with AWS and Azure. Cost structures are simple at first, but cost-saving benefits disappear as demand grows. When you use more services and more resources, pricing becomes confusing – especially if you operate without any governance on how users consume resources.
- Time of education for IT/finding specialized workers. If your developers are familiar with cloud providers and resources, education is easy. But when there are teams of users that aren’t familiar with the complexities of navigating Azure or GCP – there may take some education and onboarding time. Consider if your DevOps teams and sysadmins are used to particular systems for managing infrastructure. AWS and GCP are fairly easy to use, but we’ve noticed that some companies are hiring Azure consultants to help with ongoing issues.
- Little room for negotiation on cloud contracts. The simplicity of cloud has generally lead to relatively fixed pricing, and so for some companies this may be frustrating. If you’re doing big business, you may naturally expect to get some form of reduction in pricing.
- Increased reliance on 3rd parties to run your IT – for many enterprises this could be considered a deal-breaker. One reason is security – if you host financial or healthcare data, you may want complete ownership of infrastructure from top to bottom. Though: each cloud provider does have options for dedicated infrastructure, and Azure offers Azure Government. But if you look at those costs for said dedicated infrastructure, then you may be better off hosting your own servers.
- Uptime concerns – While all major cloud providers have incredible track records on uptime for servers, there is always a growing concern if that because your infrastructure isn’t hosted in your data centers and you lose a connection, users may suffer or your company will be unable to do business thanks to a third party.
We hope you find this information useful to calculate cloud ROI.
There is one last thing to mention: the ‘ease’ of using cloud resources can lead to costs getting out of control and infrastructure abuse by end users (i.e. teams spinning up servers and forgetting about them, too many users having access to accounts), so the upfront work by administrators to define best practices and policies is important. If you want a case by case view on how cloud can save you money or improve productivity, many enterprises we spoke with had two strategies. One was to mandate that particular business units move into the cloud at a fixed date to track results. Another was that a set number of applications be moved onto cloud infrastructure at a time. Either way, with incremental transitions, you are able to track cost and efficiency.