With all the marketing around hyperscale cloud, you’d assume that adopting it would be easy and simple. And the irony is that it once was. When first brought to market, hyperscalers like AWS, Azure and GCP wanted to ensure their services were straightforward. But as time has gone on, these solutions have become much more complex – to the point that, often, specialist training is needed right from the start.
In recent years, this increasing complexity has started to influence businesses to reconsider their hyperscale cloud usage in favour of alternative infrastructure solutions like colocation and bare metal hosting. In fact, 94% of large US organisations claim to have worked on some sort of cloud repatriation project in the last three years.
Falling short of expectations
There’s an assumption that hyperscale cloud has been built to meet one particular need. In reality, more often a need is being created around these products. As many organisations are finding, modern hyperscale cloud platforms are not straightforward compute environments. There are a huge number of products on offer ranging from platform-as-a-service to software-as-a-service and database-as-a-service.
This isn’t in and of itself a negative, but it does make it harder for the average hyperscale cloud customer to remain fully on top of their environment. And, let’s face it, most of the time becoming an expert in every single product and service just isn’t feasible.
It also means that businesses are more likely to build out their infrastructure around these (often proprietary) products and services which makes it much harder to get out when the bills get too high or the service falls short of initial expectations.
And, of course, nothing good ever comes for free.
In the early stages of building or scaling a business, particularly when VC funded or bootstrapped, free credits from a hyperscaler could be a crucial lifeline to help get IT infrastructure in place. But when the free credits dry up, your bills can get very expensive, very quickly.
And if your architecture is designed around a specific hyperscaler’s product or solution, then migrating becomes even more challenging. It’s these customers that end up locked-in, unable to escape the extortionate bills.
Free credits are all well and good, but every organisation needs to be wary of building their IT infrastructure in a way that creates a dependency on one particular hyperscale platform.
Another common misconception is that big names will offer exceptional support. In my experience, this is where hyperscalers let themselves down the most and it’s not uncommon, either. Any hyperscale cloud customer will have experienced disappointment with support services during their partnership.
Many organisations find they rarely get the attention for things that matter to their business, but are not critical for their cloud provider. So, once a business has progressed beyond a hyperscaler’s sales team, day to day technical support can be hard to come by without having to pay a significant premium on top of existing spend.
When hyperscale is the right choice
All this is not to say that hyperscale cloud is never the right choice. In many circumstances, it is.
For new businesses with tight budgets, investing in hardware is rarely a viable option. And to these businesses I’d say: go and take the free credits. Just avoid building your architecture in a way that ties you to the provider in question. This way, you’ll have an escape route when bills start rolling in which will make it far easier to migrate away to a more cost-effective solution without needing to completely re-engineer and re-architect.
Similarly, if your business has unpredictable scaling requirements, hyperscale cloud may be the way forward. Take the likes of Netflix, for example. With volatile resource requirements, hyperscale cloud is a solid option to manage near-instant spikes in demand. But for many other businesses, resources need to be exceptionally unpredictable for hyperscale cloud to be the most cost-effective option. In most cases, hyperscale cloud works best as part of a hybrid infrastructure environment combined with bare metal, or colocation.
Words of encouragement
Navigating hyperscale cloud is complex. Discovering that the reality of these environments can be wildly different from initial expectations is common.
The good news is that there are many other infrastructure alternatives out there. Bare metal hosting is one option, colocation and on-prem setups can also be effective. And, ultimately, a business doesn’t have to (and probably shouldn’t) rely entirely on one single compute type. Just consider what type of infrastructure is right for each location. It’s unlikely to be the same everywhere.
Want to learn more about cybersecurity and the cloud from industry leaders? Check out Cyber Security & Cloud Expo taking place in Amsterdam, California, and London.
Explore other upcoming enterprise technology events and webinars powered by TechForge here.
