What’s not to love about the pay-as-you-go model of cloud computing? Your service price is distributed over a longer period of time, and your customers “repurchase” every month, rather than use a solution they purchased once for years. There are few downsides – except for the added cost of hosting. Once the hosting grows to the point where it becomes a significant cost, the idea of a private cloud sparks. “Why are we paying someone else instead of setting up our own cloud?” Before you act on it, I urge you to rethink the whole cloud computing business model.
Future cloud providers often have a hard time seeing the whole picture before it’s too late. They are so in love with the idea of having all the benefits of a public cloud in a local (private) environment, that they simply overlook the pitfalls and hidden costs of cloud.
Seeing the big picture is paramount in the cloud business. What does a company need in order to run and operate its own cloud or sell cloud-hosted services to other companies? Someone has to build the data center, someone needs to run and manage it, and they will also need someone who will sell its services (data centers should be making money, right?). The owner of the data center has to fork over heaps of money in advance, while their competitors continue to use the pay-as-you-go model. When building your own cloud, you have to understand that in addition to the significant investment you’ll have to make in the beginning (since you want some scalability in your cloud environment), you’ll be faced with quite a few regular monthly costs (electricity, personnel, connectivity etc). In addition to the costs you can foresee, you should also plan for the unexpected. You, as the owner/provider, will be the one that bears all the risk should things go sideways.
Make it a million
So how much does it cost to play in the cloud service provider space – what’s the total cost of a new data center? A decent data center will cost you about a million (euros, pounds or dollars). This is the raw cost of equipment, if you plan to have any economies of scale. Once the data center is set up, you need to power and cool it, which adds a significant electricity cost. A running piece of hardware then needs the operational support. As you might remember from my earlier writings, IT personnel is the most expensive part of any IT solution. You, as the provider, will have to deal with all the complex management and maintenance of the data center, and that requires lots of technical and expert knowledge, which doesn’t come cheap.
A trap for system integrators
Many system integrators fall into a self-inflicted data center trap. Mesmerised by the “cents per minute” or “euros per hour”, they assume a data center is the best way to expand their business model, and that it will help to improve their margins. System integrators often have some in-house tech knowledge, but the business part is often lacking.
The go-to-market strategy with the cloud is very much different from the box-moving business model. The sales cycle of selling cloud solutions and services is proving to be very long, and companies tend to postpone the decision on moving their data and business to the cloud again and again. This is primarily because moving your business to the cloud is very much a one-way decision - it’s much easier to decide to move into the cloud than to get out of it, and companies know this. So in order to sell a cloud solution, you will also need skilled sales people with strong technical and business knowledge. They need to have a profound understanding know of the way clients do business and how cloud solutions are going to help them. Until this is established, every deal stalls in (admittedly enthusiastic) consideration stages.
System integrators who see cloud as little more than an add-on offering are in for an even bigger surprise. Not doing things the right way or cutting corners in the wrong places can be costly. Just adding some blade servers and storage to the existing server environment won’t make you a cloud provider. There will be no real scalability, and the cost benefit will be mediocre at best.
Some providers try to lure clients in by promising them that they can deliver just about anything, even mirror their environment in the cloud. While this is certainly doable, it is also costly. The true cost of these promises becomes evident later on – when the provider figures out that instead of standardized solutions, which provide cost efficiencies, they need to deal with tons of exceptions and legacy systems which cause nightmares with data center management (and drive costs through the roof).
Do the math beforehand
Selling cloud services requires an enormous effort. Most companies opt for the competitive position of low pricing with some big promises. These promises might include cost of migration (data, applications, processes). In practice, migrations prove to be incredibly costly and can easily reach price tags exceeding 100,000 (again – euros, pounds, or dollars). Offering these services to the client free of charge in order to get the business has caused more than a few cloud service providers to scale back after reality set in. Some of those who stubbornly convinced themselves that the next customer migration will be different and that it’ll fit their idealized scenarios never recovered their losses and shut down.
The combined costs mentioned above mean that most cloud providers need to keep their customers for at least two years to break even. Only after two years do they start making any money, and that assumes they made all the right decisions along the way. The investment is monumental; if you want to try your hand at cloud hosting, you need lots of customers, you need to scale your growth very well, provide standardized solutions, hire expert personnel, and a bit of luck.
So why do companies insist on building data centers if there are so many pitfalls along the way? Not seeing the big picture is a costly mistake, and even among those who understand the total cost of such a strategy, many believe that keeping their customers for the long run (ideally, forever) will eventually pay off everything they invested. Is this naive?
I’m sure that the future will bring technical development which will make data center hosting and cloud provisioning more accessible. Still, today, the cloud business is only truly accessible to companies who are experts in the field. Others are learning this lesson the hard way.