An abundance of assumptions have appeared in the discourse around public cloud procurement – that public cloud is always cheaper than private deployments, that cloud services are essentially an undifferentiated commodity purchase, that migration to public cloud is a simple process, or that the best deal can be found by sourcing from the market leading, juggernaut provider. But what if it was possible to challenge some of these assumptions with real data on pricing for various deployment options?
This is a task that the 451 Research Group has set itself with its Cloud Price Index, a study of pricing for “a basket of cloud services” that covers requirements for a multi-service, three-tier cloud application made up of Linux virtual machines, object storage, block storage, relational database, NoSQL database, load balancing, access control lists and snapshot backup in a resilient architecture. This collection of services may not apply to all end user circumstances – an ecommerce business may have different priorities than a manufacturer and hence different cost criteria – however, it works to set a baseline against which price fluctuations can be measured. “The idea came from inflation, which is measured against the consumer price index – a basket of goods that most people buy every quarter,” explained Owen Rogers, research director of 451 Research’s Digital Economics Unit, and project lead on the Cloud Price Index. “The basket of cloud services is really just a benchmark basket that we can use to track how the actual market is changing over time. There was a lot of hype in the market a couple of years back about the price of cloud coming down, about commoditization, but I had a hunch that if the price of virtual machines might be coming down, there were a lot of other services that were changing [in other ways].”
Launched a year ago, the Cloud Price Index has produced a number of intriguing findings. Based on an initial analysis of pricing data, 451 researchers concluded that contrary to popular perceptions it is not always the case that a cloud service is a service, is a service, rather that there are major price differences between providers, and that major gaps appear in pricing for services “up the stack” which may serve as a source of vendor differentiation. Later in the year, 451 analysts tackled private infrastructure, comparing the costs to build private cloud based on proprietary technology, commercial OpenStack distributions, and DIY OpenStack services – adding in perspectives on public cloud and managed hosted private cloud – to conclude that TCO is highest in the case of DIY OpenStack services, due in large part to the need, and scarcity of OpenStack engineers.
This month, the Cloud Price Index has generated yet another thought provoking finding – that cloud buyers can find substantial savings by shopping around for the cheapest services for each component of their web application. According the latest quarterly Index update, buyers who make long term commitments but mix services such as compute and storage from multiple providers, rather than source from one provider, can save 58 percent in the case of small applications, and up to 74 percent in the case of a large application – for a total cost saving of $23,000 per application over a three year period. This analysis is outlined in the figure below, which compares pricing for a small and large “basket of services” sourced from a single or multiple providers. “On-demand” is standard pricing, while “Best case” takes into account commitment, volume or other discounts, which Rogers noted might actually better represent the fees an enterprise would pay if they were to negotiate and commit to a provider.
As the figure shows, the cheapest procurement practice in all four cases is to source services from multiple providers. But is the cheapest option necessarily the best – or even the most realistic from a TCO perspective? ‘Shopping’ entails another assumption – that cloud migration presents little obstacle to optimizing procurement based on price. However, data and application portability have proved a tricky business for many cloud adopters, and particularly challenging when applications are coded directly to a provider’s platform – requiring considerable effort to manage various cloud integrations, APIs and provider onboarding processes and billing systems. “If you are an end user,” Rogers observed, “you should look to a single provider who is integrating all these services, and making them easy to use, which would include porting data to support an exit strategy. If you are an end user and not getting that integrated service and that benefit from using a single provider, then perhaps that benchmark can be used to assess the value of individual services. Knowing you can save that much gives users some ability to negotiate with their provider to ensure they are getting the appropriate benefit and value.”
Beyond a tool for the end user to hold service providers accountable for price and/or service benefit, the Cloud Price Index findings on sourcing cloud from multiple providers presents an ideal opportunity to the ‘cloud brokers’ looking to prove out their value in an evolving marketplace that is skewed towards direct sales. If managing the multi-vendor cloud proposition is complex for the end-user, providing best-of-breed solutions is a long standing value prop of the channel which may be translated to the cloud marketplace. While Rogers noted that “providers could offer decision engines to help enterprises choose the best-value mix of services to suit their needs, and then integrate and manage these services as a managed offering,” cloud channel players could also use pricing analysis to help build their own vendor agnostic decision engines to fine tune advice on various provider offerings, while creating a platform that could provide the basis for mixing and matching different services. “Brokers could offer a single portal, they could do API translation, and offer a single support function, while offering one throat to choke in the management of multiple providers. In many ways, they can offer the best of both worlds: providing the cost benefit of using multiple providers, while removing the issues associated with integrating various platforms,” Rogers added. “Anyone who is already doing integration of systems is in a natural position to be this integrator of clouds.”