For the past few years, Infrastructure as a Service (IaaS) providers have been at the mercy of big brand price wars. In October 2014, Google dropped its cloud instance prices by 10%; then in November, Amazon countered with a new tiered pricing structure promising lower costs for customers. As Google, Amazon and Microsoft slash costs on a seemingly constant basis, customers must be living the good life – right? You would think that was the case, but the reality is that these price cuts are cloaking the fact that the majority of companies are already paying twice as much as they should…
The Big Cloud Cover-Up
Cloud price wars are a mere distraction from the fact servers are typically only 51% utilised over the full 24/7 cycle; meaning users are paying nearly twice what they should be. Despite the fact businesses have embraced cloud, many have moved their on-premise bad habits along with them into the cloud. In the on-premise world, businesses had to over-provision in terms of the numbers of servers they would purchase as they needed to run with additional headroom in order to handle peaks in demand.
When moving to the cloud, the same thinking has been applied. We conducted a survey of 200 CIOs in September last year which found this to be the case; 90% of those surveyed said they saw over-provisioning as a ‘necessary evil’ in order to protect performance and ensure sudden spikes in demand are handled.
While cloud services are positioned to be on-demand, pay as you go, in reality they are still charged by capacity rather than actual usage. Aside from deploying complex software to automate the process, companies have to provision capacity manually and select from a restricted number of instance sizes. This creates waste, as the response time is slow and scaling steps are large, so concerns over performance and availability force companies into paying for a buffer of excess space they are not actually using. This block billing approach to cloud server provisioning is increasing the cost of cloud infrastructure.
But simply, businesses should not have to choose from a fixed menu of instance sizes, pay for services at a scale that they ‘might’ need, or buy capacity that they are not using: yet this is what most companies are currently doing as they do not see an alternative. Yet new technological developments are enabling a new way of working that will mean pay by capacity billing will be a thinking of the past and cloud price wars will become irrelevant.
Next Generation Containers
Container technology used to be viewed as a low-end product, criticised for being inflexible with lower performance. However, recent updates to the Linux kernel have meant that containers are having a revival. Over the past year or two we have seen players, such as Docker, making headlines with their innovative use of containers to increase application portability. We are now seeing the revolution spring up to disrupt the cloud infrastructure market.
Cloud providers have brought new products to market, like elastic auto-scaling servers using containers. For customers, containers spell savings. By enabling providers to scale on demand, servers can precisely resize according to consumption, rather than increasing in fixed increments determined by the instance sizes on offer.
Having the ability to dynamically scale opens up completely new billing models, as providers can start to bill by usage, rather than predicted capacity. Containers provide companies with an auto-scaling, self-managing, elastic infrastructure; completely removing the need for manual management or provisioning. By removing the need to provision additional servers during peak periods or configure software to manage this provisioning process, organisations can simply turn on the service and forget about it.
Each server elastically and instantly scales up and down when more or less capacity is needed to meet users’ requirements. This means that high performance is always available, but costs are kept to a minimum at all times, as billing is based on actual usage, rather than server size or capacity. As a result, companies only have to pay for what they actually use, right down to the last MB.
The gig is up
While providers such as Google and Amazon use containers internally to help streamline their businesses, they do not use them to deliver more favourable pricing for customers – instead, they keep distracting them with price cuts that are essentially meaningless, when you consider that people are already paying twice as much as they should. My question to organisations is: why would you ever choose to pay for capacity you aren’t using, sacrifice performance, and eat up your systems administrator’s time, when you don’t need to? Soon companies will be asking themselves the same thing, which will mark the end of capacity-based billing.