Delta: Cost comparisons for the cloud hyperscalers

Few organisations see cloud primarily as a cost-saver these days, but there are some notable differences between the big providers

Growth in public cloud continues to be measured in double figures as companies move commodity data and applications out of their own data centres and replace hardware-bound applications with cloud-based ones at a rate accelerated by the pandemic.

The leading cloud platforms show no signs of slowing down, constantly adding new services and buying up smaller competitors in key areas like AI. However, the growth of the "hyperscalers" in the public arena is occurring in tandem with some other market trends. One is towards data and application portability and the nascent "multi-cloud" or best-in-breed approach to public cloud services, and another is the growth of hybrid cloud, which is becoming a strategy rather than an accidental halfway-house.

One of the original cloud drivers was its apparent potential for reducing costs, and certainly some companies can save a great deal by rationalising their operations in the cloud. What's more, cost savings are always an easy story to tell the board.

Click here to access the cloud research

However, while some may genuinely be able to cut expenditure by moving data centre operations into the cloud, this is by no means guaranteed and any savings depend very much on the use case. Dedicated high-performance computing, for example, is likely to be much more expensive when deployed in the cloud as an IT leader working in scientific research explained:

"Once we added it all up - licences, use of infrastructure as a service, storage, RAM, CPU -the cost was about double for the same level of service that we get internally".

Cost comparisons between on-premises and cloud can become fiendishly complicated, with numerous factors to take into consideration including software licensing, hardware depreciation, energy costs, skills and the cost of doing nothing. And it's easy to underestimate the ongoing costs of cloud services as usage ramps up, as it typically does.

While vendors typically make it easy and cheap for customers to get on board, many of our respondents complained of facing additional costs when it came to adding new services. Cloud is full of price traps that can snare the unwary. For most organisations any savings will be small, if they are realised at all.

Satisfaction scores on price-related issues were all fairly low for the big cloud vendors, but not disastrously so, suggesting that organisations know what they're getting into these days.

Out of the four biggest cloud services used in the UK, Google Cloud Platform scored the highest on cost related issues, but such comparisons should be handled with care. The types of projects most likely to be running on GCP and, say, IBM Cloud may look very different.

In addition, basic services such as block storage may be very cheap, whereas more specialised options can come with a higher price tag, or with pricing steps that can easily catch out the unwary.

The two largest players, AWS and Azure, both scored the lowest on portability. Despite noises favourable to the multi-cloud idea, incumbents still make it far easier to put your data and applications on their clouds than to remove them. High data egress charges are one of the most common complaints we hear about cloud services.

This is a sample of the information in Delta's Cloud IaaS/PaaS research, which contains all the latest information on the market, vendors and tools - including pricing information and feedback from customers who have used the solutions in anger. Click here to ask about access.