There’s a reason why hyperscalers like Amazon Web Services (AWS), Google Cloud Platform (GCP), and Microsoft Azure have done so well: they provide cloud compute and network infrastructure services on a large scale and they’re easily accessible. The platforms are attractive for many users and development start ups because of their almost unlimited service options and resource availability.
They make starting and growing a business so easy that it’s hard to resist the benefits these sizeable public cloud providers can offer.
How easy is it to switch?
But while they all offer similar services, every hyperscaler is built differently. The services they offer are proprietary which means developers can often find it hard to change platform. It’s not quite as drastic as “vendor lock-in”, it’s more a case of subjecting themselves to ‘high switching costs’. Businesses that use a hyperscaler’s proprietary services may not realize just how high the costs could be to switch to another hosting platform or provider – not to mention the time and effort involved.
People would be understandably reticent to sign up with a hyperscaler if they knew that leaving could be much harder than they imagined. While this isn’t really an issue with basic services, the threat of lock-in becomes much more noticeable when it comes to more advanced services.
For example, AWS S3 object storage is a very convenient service to store new data at what appears to be a low cost. But for many organizations with high volumes of data the AWS pricing model makes it unaffordable to migrate away from that service.
How expensive is it to scale?
When a business grows in popularity, scaling of the underlying resources and increased usage can result in unforeseen invoices due to the intrinsic pay-per-use models used by hyperscaler platforms. For many customers, this means they may need to hire dedicated consultants to manage their hyperscale infrastructure usage and prevent costs from skyrocketing. Networking costs, for example, can quickly become more than expected with hyperscalers.
Is pay-per-use a solution?
Pay-per-use models are widespread with hyperscalers and they can seem very attractive in cases where businesses only use what they need and when they need it. But it’s not quite as simple as that. Businesses can underestimate how frequently they might need to run an application. In many cases, an application might be required 24/7 or it can’t be scaled down when the load isn’t high enough. That’s when pay-per-use models can suddenly become very expensive.
What about pay-per-reservation?
For many organizations the solution might be to opt for infrastructure as a service, using standard (open source) software solutions to run services instead of ready-to-use proprietary services on hyperscaler platforms. This allows them to retain a better grip on their flexibility and independence. Running software on standard infrastructure services at ‘fixed’ fees (pay-per-reservation) with a hosting provider typically results in more predictable invoices at the end of the month. In many cases, companies can benefit from more resources or higher performance at lower costs.
A multi-cloud strategy could be the solution
Hyperscalers are making big investments in innovation but they’re not doing it for nothing. Sooner or later, they will want to earn it back. Given the problems of technical lock-in and the high cost of migrating from a proprietary platform to an alternative, businesses may want to spend more time identifying the risks before they embrace hyperscaler services fully.
The benefits provided by hyperscalers can seem very attractive, but it is worth considering whether adopting a multi-cloud strategy might provide more flexibility and independence now and into the future. A multi-cloud strategy enables businesses to interconnect their environments with multiple cloud providers and run workloads on the most efficient and cost-effective platforms as demands, needs and costs dictate.