Cloud Isn’t Everything: The Benefits of On-Site Servers
In this article, I want to dispel some of the common myths about on-premises servers and explore the benefits of having them. Contrary to popular belief, running your own servers on-site is a viable infrastructure option for your organization. There are a lot of misconceptions floating around online about cloud computing and on-premises servers that are very deceptive.
This article might ruffle a few feathers in the tech industry, but this topic needs to be discussed. Too many businesses are spending far more money than they should be on technology, driving up their monthly operating expenses.
Table of Contents
- CAPEX vs. OPEX Spending
- Myth #1: Purchasing physical servers is wasteful spending; it’s too easy to purchase more/less than needed
- Myth #2: On-premises infrastructure is obsolete
- Myth #3: The cloud is cheaper than on-premises servers
- Myth #4: The cloud is more reliable than on-premises servers
- Myth #5: CAPEX spending on technology is senseless; your purchases depreciate over time!
- Myth #6: There’s a cloud app we really like. If we install on-premises servers, we can’t use the cloud anymore!
- Conclusion
CAPEX vs. OPEX spending
In this day and age, it seems like everything is sold as a subscription of some kind. It’s becoming increasingly less common to buy something once and own it forever.
Let’s set aside your business expenses for a second – think about just your personal expenses for entertainment. If you’re like most people, you probably have several monthly subscriptions to different music and video streaming services. Instead of buying the media once and owning it forever, you pay an ongoing subscription fee forever, and if you stop paying, you lose access to it. Have a look at your most recent credit card statement. If you add up all of the monthly charges for your streaming services, I bet you’re paying close to $100 per month or more – and that’s just on entertainment!
Before we get into the myths below, I want to briefly visit CAPEX (capital expense) spending and contrast that with OPEX (operational expense) spending. CAPEX is when your business purchases something outright as a one-time transaction. The upfront cost is a lump sum that is typically larger than the average everyday purchase, but the initial high price you paid gradually amortizes over the length of ownership until a definite ROI (return on investment) is reached. In this way, CAPEX spending is best thought of as investment spending. OPEX spending represents perpetual costs to your business. OPEX can take the form of leases, rentals, ongoing required purchases, and subscriptions.
For example, when you buy a new printer for the office, that’s a CAPEX. But the ink you purchase for this printer is an OPEX because it’s an ongoing expense, similar to your electric bill. OPEX is the same financial model that cloud computing falls into. In fact, cloud computing is best thought of as another utility bill. To sum it all up – OPEX is about ongoing expenses, while CAPEX is about one-time investment purchases.
At the risk of belaboring the point, in terms of ownership, I like to think of the cloud as analogous to renting an apartment vs. buying your very own home. With an apartment, you do not build any equity in the property. Stop paying your rent, and you must leave, no matter how long you’ve lived there. A house is a larger investment, and you’ll pay a mortgage each month, but you are building equity over the life of the loan. As you pay down your mortgage each month, your ownership stake in the home increases while the lender’s ownership stake decreases. When the mortgage is fully paid off, the home is 100% yours.
When the cloud first gained traction, one of the main draws was this idea of favoring OPEX spending over CAPEX spending. The reasoning is that it’s less expensive to simply pay-as-you-go every month for only the resources you need. The notion of elasticity also caught on, meaning you can scale cloud resources up or down as needed, making the end cost to you more predictable. In fact, depending on the cloud platform, this scaling can even be automated so that more resources automatically spin up during times of peak usage, and spin back down when the peak has ended. Sounds great on paper, but in practice this is fantasy. It seldom, if ever, pans out that way in the real world.
Next, let’s look at some of the most common arguments made for abandoning on-premises servers and going all-in on the cloud. These arguments are myths, but unfortunately they are widely believed, because they contain what I like to call “half truths”, which makes the myth harder to spot.
Myth #1: Purchasing physical servers is wasteful spending
The idea of this myth is that by purchasing your own hardware, you will either be under-provisioning or over-provisioning. Over-provisioning is when you purchase a server that’s far more powerful than you need, while under-provisioning is purchasing a server that isn’t powerful enough. The myth goes that you will always over or under provision, and that there’s no way to purchase exactly what you need. You’re told that the only way to get this level of scalability is to go with the cloud.
Let’s break this myth down piece by piece:
Under-provisioning:
If you purchase a server that is not powerful enough for your workloads, this is wasteful spending. Pretty obvious, right? This is where the “half truth” aspect comes into play – this fact about under-provisioning is definitely true, but this is where the truth ends. Now let’s examine the other side of the coin.
Over-provisioning:
If you purchase a server that’s more powerful than you need, this is seen as wasteful spending because most of your server’s hardware will be underutilized. This sounds like a bad thing, but the reality is – in this day and age, over-provisioning is a desirable outcome. Allow me to explain –
Over-provisioning was only a problem back in the 1990s and early 2000s, prior to the advent of virtualization technology. Back in those days, you needed to purchase a physical server box for every server that you needed to run in your organization. This was wasteful spending because if a server had 16GB of RAM and a blazing-fast processor, but only needed 3GB of RAM and few processor cycles to run your software, that extra hardware was an unnecessary purchase.
Today, workloads can be run in virtual machines and containers. Modern on-premises servers are virtual machines that can be sized appropriately. This allows you to pack several “virtual servers” into fewer physical server boxes. With modern day virtualization, you want to purchase a more powerful server box than you need, to allow for future expansion. You may only need one or two servers today, but if you need an additional server down the road, it is a simple matter to spin up another virtual machine or container on that physical box. You no longer need to purchase new hardware every time you need more workloads.
Myth #2: On-premises infrastructure is obsolete
This is a marketing gimmick, and nothing more. “If you aren’t running 100% of your operations in the cloud, your business can’t compete in the 21st century!”
There is nothing “obsolete”, “legacy”, or “old-fashioned” about running servers on-site. On-premises infrastructure is very much still a viable deployment option, and there’s quite often a solid business case to be made for it.
Nothing would make the major cloud providers happier than to have you run 100% of your operations in their clouds. It is a perpetual source of income for them. You may also see some I.T. providers pushing 100% cloud because they earn recurring commissions on those sales. Yes, we sell cloud solutions here at SignalCircuit, and yes – we do earn recurring commissions on those sales. But we only deploy these cloud solutions when and if it’s in the best interest of the customer. We perform plenty of cloud and on-premise deployments and migrations, and happily support both infrastructure types.
Myth #3: The cloud is cheaper than on-premises servers
This can be true, depending on the individual organization. We find that the cloud is cheaper for very small startup companies that do not have a lot of upfront capital to begin with. They can’t invest in physical servers, and so it makes more financial sense to use cloud-based software for their operations.
Here’s the thing – much like your entertainment subscriptions we talked about above, most cloud software is priced per user, per month. If a startup only has three employees, and the software costs $10/user/month, the total bill is $30/month. Easy math. But as the organization starts to grow and become more established, if they want to continue using this software, they will need to purchase additional licenses to cover the additional staff they need to hire.
An organization with 20 employees using this same software would be paying $200/month for this cloud app. Sometimes there are volume discounts for larger organizations, but keep in mind this is still an ongoing operational expense – you will never own this software and must pay for it forever if you want to keep using it!
A physical server is more expensive to purchase up front, however, over time the investment will pay for itself in the savings. When you crunch the numbers, over the long run, the TCO (total cost of ownership) for on-site servers wins out in the end.
And if virtualization is being used, as discussed in Myth #1, you may not need to purchase any additional hardware at all. You may only need to purchase software or operating system licenses, depending on what role this server will play in your organization. If this new virtual server is only going to run free open-source software, you wouldn’t need to pay anything at all!
Myth #4: The cloud is more reliable than on-premises servers
This is another one of those arguments that can be true, depending on the context and circumstances.
In one sense, the cloud is more reliable in that if you lose your internet connection, your software and services up in the cloud will still be running and responding to your customers. On the other hand, if you lose your internet connection, and you’re doing everything in the cloud, your entire company’s operations will be dead in the water until service is restored. So in that sense, it’s less reliable. An on-premises server running software locally is immune to internet outages. A UPS battery backup can be added to protect it from sudden power outages too.
Another point to keep in mind is that cloud providers are not invincible to the elements. They cannot guarantee you 100% uptime because it simply isn’t realistic. They are subject to power/internet outages just like anyone else. If one of their data centers has an outage, and your servers are living in that data center, your operations will be affected by that outage.
Myth #5: CAPEX spending on technology is senseless; your purchases depreciate over time!
This is another half truth – your purchases indeed depreciate over time, but that doesn’t mean that purchasing as opposed to leasing is senseless. Unless you plan on reselling your hardware, depreciated value isn’t an issue. When was the last time you resold a used computer or other equipment from your office? Most businesses use their purchased equipment for as long as possible, to get as much useful lifetime as they can get out of it.
When the equipment has reached the end of its useful life, most businesses decommission and recycle the equipment, rather than resell or give it away. And they have the data on old servers and computers securely destroyed to prevent data breaches.
The typical refresh cycle for server hardware is 5-7 years, but if properly maintained, you can get a decade or more of use out of a server box.
Myth #6: There’s a cloud app we really like. If we install on-premises servers, we can’t use the cloud anymore!
This one isn’t even half true – it’s an outright lie! Choosing between cloud and on-premises equipment isn’t an either-or proposition; it’s either-both. Cloud and on-premises work excellently hand-in-hand. Using cloud and on-prem together, also known as hybrid infrastructure, is a very sensible approach and has been gaining a lot of steam in recent years.
We’re not suggesting that the cloud is a bad thing. On the contrary, the cloud has many wonderful benefits. The point we’re making is that you can reduce your operating expenses by reducing your cloud footprint. There are many types of workloads, and a lot of them do not need to be in the cloud. The more workloads and software you can offload from the cloud and into your own on-site environment, the lower your cloud bills will be. If there is a particular cloud app that you really like, you can certainly continue to use it alongside your on-prem infrastructure.
Conclusion
Well – this was a rather long blog post, quite possibly the largest one we’ve published! We’re always posting technical information like this with the hope that you are gaining business value from it, and greatly appreciate your feedback. Hopefully at least some of this own vs. rent material has resonated with you – the cloud is a massive giant, and it can be intimidating to even consider using anything other than what “everyone else is using”.
I want to hammer home the point that this post is not intended to discourage use of the cloud. The cloud is not a bad thing, quite the opposite in fact. The cloud will always have its place in the industry. There are many benefits to using cloud services. What we’re saying is that it does not supplant on-premises servers. On-premises infrastructures will always have a place in the industry too. They are not obsolete and not going anywhere! We typically aim for hybrid-type deployments where a mix of cloud and on-prem servers are used, so that you get the best of both worlds as well as reduced operating costs.
At SignalCircuit, we offer a free initial consultation, so if there is anything about cloud computing vs. on-premises servers you’d like to learn more about, or want to discuss any other business technical needs, feel free to give us a call!
