One of the most significant barriers to cloud adoption is perceived costs - understanding the cost of cloud services is a longstanding question mark in the minds of businesses everywhere. To the extent that many organisations are dismissing public cloud as a financially viable option from the offset, while other businesses that have already adopted the cloud find their monthly bills spiraling out of control.
So just why is the cloud so expensive?
The simple answer is, it isn’t. Spending most of my time in front of customers and seeing these objections, I usually find that digging a little deeper brings it back to one of the following reasons:
Organisations often turn to online pricing calculators provided by the leading public cloud platforms for an easy way to estimate your monthly cloud costs. These calculators can be extremely misleading – and often lead people into thinking cloud is going to be 2,3 or even 4 times the cost it actually will.
Think about this, would you use list pricing to buy an Enterprise Storage Array? Would you trust yourself to spec the right equipment without engaging said storage vendor’s specialists?
About 10 years ago, we all marvelled at the birth of virtualisation, wowing at the technology seminars where demonstrations of virtual machines seamlessly transitioning between 2 physical servers seemed like some form of trickery that could never be real. Back then this was amazing, and more importantly it was very, very different.
You knew you needed help with virtualisation, you couldn’t possibly try and get started on your own and it was so ‘mystical’, you almost needed the comfort blanket of an expert to blame if it didn’t work out.
With cloud however, the fundamentals are far less concerning. Most people understand that both AWS and Azure are built upon industry standard hypervisors and IaaS workloads are pretty much just ‘virtual machines’. Also it only costs about £15 a month to turn one on and have a go, so the financial impact of doing something wrong isn’t going to get you fired.
So on that basis we can all just crack on and scope what the solution would cost, right?
This predetermined comfort with cloud is where the big problems stem from – asking a virtualisation engineer to go and scope a public cloud environment (which is seeming just a virtualisation platform consumed on a pay as you go basis) is common practice, and not really that illogical on the face it. And to be honest, it would probably be fine was it not for the virtualisation engineer’s mindset.
The Virtualisation Engineers Mindset
Virtualisation engineers have several ‘laws’ they tend to abide by (or at least I did):
Overprovision for stability – driven by the software vendors ‘minimum requirements’ of the last 10 years, and that fact he didn’t want to be waking up at 3.00am on a Sunday morning.
Architect for the masses – designing infrastructure platforms that will host all manner of applications and meet the diverse and often extremely high Service Level Targets for the next 3/5 years.
Utilise traditional availability models – Virtual Machine Level Replication, High Availability Hypervisor Clusters and Centralised Storage to name a few.
These ‘laws,’ as great as they are, have been at making us look good for the last 10 years are absolute no-go’s in the cloud world – not really from a technical perspective but purely from the economics and financial viability of it.
Cloud by its very nature is a pay-as-you-go service, which means you need to be diligent about how you consume it. IT teams are not used to worrying about this, yet it is absolutely crucial to avoid being stung by a painful bill. Organisations have a responsibility to pore over the monthly costings in fine detail, but this is often easier said than done when the billing usually makes little to no sense.
Further to this, many applications in their current form aren't designed to run efficiently in a public cloud platform, which ultimately means you can’t expect to just lift and shift your apps. To avoid paying more than you need to, carefully consider which applications need to be either re-platformed, or at least optimised to run efficiently in a cloud environment – right down to when the service should be live (turned on), where deduplication can be used, how data will be protected – ultimately any cost centers that can be manipulated for efficiency.
Trying to scope a public cloud solution any other way, will without doubt, make it appear eye wateringly expensive in comparison to traditional infrastructure platforms.
To plan and accurately price a public cloud provision or migration takes a lot of time and effort – it’s arguably the most important element of every cloud project, and gets missed way too often.
If you’re reading this and thinking your public cloud is well optimised and costing what it should – you're probably wrong. As Head of Public Cloud at ANS, I probably spend almost half of my time helping organisations that are already in the cloud become more efficient, and no matter how lean they think they are, we nearly always find ways to reduce the platform spend by anywhere from 20 – 50%.
Our free, no obligation ANS Check Service helps you answer some of the key questions users of Public Cloud often have including; how secure is my cloud environment, is the architecture in line with best practice and why is the cost more than I thought?
To find out more about our cloud check service, click here
Posted by Joe Wolski