5 Reasons Why The Cloud Approach Won’t Work For Your Company

In my previous articles of my mini-series about the cloud, „The 3 Most Important Reasons Why The Cloud Will Win“, and „The Infrastructure Dilemma“, I praised the long list of benefits when a company decides to go fully into the cloud, changing its system architecture and IT processes to leverage the full potential.

This time, I will give you, from my experience with my clients over the last years, the most important reasons why they think their company can’t move into the cloud — luckily, most of the THOUGHT they can’t move :-)

1. „We Are Different Than The Others“

Oh well, how many times did I hear that….and not only when talking about moving into the cloud, it’s one of the main reasons why NOT to change anything in the system architecture.

To my experience, such companies just want to be different for being different, without added value to the customer, and to differentiate them from competitors. While the later might be even true on the business side of the company, it’s almost never true for the IT and the processes they need to support. Especially those processes which do not create revenue or have no customer impact at all.

In case you hear that, let’s hope you just asked the wrong person. Maybe moving up one level in the hierarchy makes more sense. People working directly in Infrastructure or IT (and not having some management position) tend to ignore or misunderstand the benefits of the cloud, mainly because it’s either out of their comfort zone and they are happy about as it is (but usually still complain), or they fear they can lose their job when the company moves to the cloud.

2. „We Have Monolith ABC That’s Why It Makes No Sense To Move To The Cloud“

„Aha. And this monolith is the only software you are running in your datacenter? Oh, there is a middleware, and a service layer, and 30 other software components, too?“

Guess you know where this heading towards…

If there is a monolith (or even more than one), you have two valid options to move to the cloud:

1. If the monolith can’t be moved, and is surrounded by systems where latencies of 10–20ms are not a problem at all, move everything else to the cloud and leave the monolith where it is. The extra effort for moving all the other stuff to the cloud will pay off soon.
2. If there is a chance to move the monolith, too, to the cloud, move everything there. Even if hosting the monolith with the cloud provider is more expensive, the cost reduction for all the other systems pays this difference easily.

3. „We Need Tight Integration“

Well, which system doesn’t? I’ve seen some cases where there is no way to move existing components to the cloud, basically in these two scenarios:

1. Your system landscape contains (non-IP-based) signaling boxes, e.g. in a TelCo scenario, which needs a totally different infrastructure than a cloud provider can give you.
2. Your system landscape contains components which need to be interconnected with such signaling boxes or networks for extreme throughput, e.g. the billing system in a TelCo which receives hundreds of millions of events from Core network which need to be rated etc. and provide feedback if a customer still has sufficient fund for the call he requested, etc.

The good thing, though, is that somewhere in a huge system landscape with one or more of the above components, there is a boundary where latencies of 10 or 20ms are no longer an issue. For instance, while the billing or mediation components need very fast connections to Core network components, they need less capable connections to the ERP system or CRM system. So from this boundary on you can move everything to the cloud.

4. We Need Control Over Our Systems

Sounds like a no-brainer, right? Who doesn’t want to have control over his systems.

But if you hear that, start asking what „control“ is in this context. I’ve never got a satisfying answer to my questions. I would assume this is due to there is none. What is seen as „control“ is an understandable, but nevertheless false feeling of touching things to be in control. It can be very satisfying to walk through the datacenter, seeing all the blinking boxes and cables, but this doesn’t mean you have everything, or anything, under your control.

Control is not established via physical access to hardware, it is established via well-designed processes, documentation and proper knowledge. This usually leads to automated processing of service requests and demands. And this is the perfect point where you can go to the cloud. It doesn’t matter if you run your deployment and environment scripts on the machine two floors below you, or at the datacenter in the same town you work, or 200 miles aways in the cloud.

If you need to access the physical hardware frequently, you have a problem you should try to solve. If everything would be running perfectly, there is no difference between a cloud provider or in-house hosting, the infrastructure should work! If something fails, and this can happen, at least it’s not your turn to fix it when everything is in the cloud.

If you get stuck at this point with some IT guys, let them draw down how they think their perfect in-house datacenter should work. The will add lots of automation, so that they do not need to deal with every single request for new instances, firewall rules, etc. They will add virtualization. In a modern architecture, they will think about Kubernetes, or at least some Dockerization, or some service management (gateway, config, registry). What they don’t know when starting this, the more they think about how it should be, the more they move into what the cloud provides out of the box….

5. „We Have Higher Security Requirements Than The Cloud Can Provide“

Ok, this can be a valid reason, if:

• your company is part of the vital infrastructure of a country, and you are not allowed to move the infrastructure to another country
• you’re working for the government, or an organization owned by the government, which is (at least in Europe) always seen as „critical for the infrastructure“

But even if your work falls into the two possible scenarios above, there are lots of systems in such organizations which do not fall under specific laws forbidding to move to other countries, and these parts could be hosted much much cheaper in the cloud.

For all other cases, I doubt that an in-house datacenter can be more secure than one of the larger cloud providers.

Ask yourself the following question: who is more likely to be able to cope with DoS attacks, hacking etc.: a company with 3–4 people dedicated to security, or a company which has not only hundreds of people working on security and detecting fraudulent behavior NOW, and hundreds designing the system and pushing it to be even more secure on a daily base? I think the answer is clear.

PS: Until now, you might have recognized that the title of this post is misleading — I can’t tell you any reason to not go the cloud, except there are regulatory boundaries or laws from preventing you to do so.