For UK education, private clouds may make economic sense
Advocates of the public cloud tend to dismiss private cloud deployments within individual companies, arguing that they are too small to deliver the cost or flexibility advantages of the public cloud. But recent developments in the UK refute that argument. The University Modernization Fund‘s Cloud Pilot, which aims to deliver diverse private computing capabilities to every UK university while saving money, has just announced its pricing, which compares extremely favorably to current public cloud offerings.
Rather than use the combined purchasing power of the UK’s universities to negotiate discounted prices from an existing cloud provider — Rackspace, Amazon, etc. — the funding agencies chose to conduct the pilot using Eduserv, a UK-based nonprofit that provides a range of services (managed hosting, access management, collaborative license negotiation) to government and universities.
What the numbers say
The UMF Cloud Pilot includes a range of virtual machine and data storage products based on VMware’s vCloud and the open-source OpenStack project. Data storage is cheap: 6.7p (around $0.10, although the service is only sold and priced in the UK) per GB per month. Broadly comparable offerings from other cloud providers cost, approximately: 5.5p per GB per month (Flexiscale), 11p per GB per month (Rackspace) and 9p per GB per month (Amazon S3’s Dublin data center).
Virtual machines for the UMF Cloud Pilot are also reasonably cheap: A single vCPU Linux virtual machine, with 1 GB RAM, costs 2.4p per hour. Its Windows equivalent costs 2.8p per hour, and a loosely equivalent OpenStack Compute “micro” instance is 1.4p running Linux and 1.7p running Windows.
Comparisons with other companies’ offerings are difficult and inexact, but Rackspace charges 4p per hour for a Linux server with the same 1 GB RAM and 5.2p per hour for Windows. Amazon’s equivalent machine falls somewhere between a Micro instance and a Small instance, with these costing 2–6p per hour for Linux, and 2–8p per hour for Windows. Both Amazon and Eduserv further discount prices for “reserved instances” that are allocated in advance, cost-effectively meeting the requirements of predictable computing loads over the longer term.
Unlike its commercial competitors, which can see significant margins, Eduserv is more likely to be delivering its service at (or near) cost. It may even be subsidized by the funding agencies driving the wider UMF program. In that context, its prices look less impressive. Without substantial growth to generate economies of scale, Eduserv has no room to further cut prices. Its competitors do.
However, Eduserv’s Cloud Pilot has one significant advantage: a direct connection to the UK’s high-speed academic network JANET. All of its potential customers are also connected to JANET, meaning that data transfer (both in and out) is free. When it comes to especially large research data sets, that could quickly become a significant differentiator. An ongoing consultation into an arcane piece of tax legislation known as a “cost sharing exemption” may also mean that Eduserv is able to waive the VAT (20 percent) that it would otherwise have to charge customers, a further advantage over its competitors.
More than a price tag
For all of this, Eduserv is certainly to be commended. But a sustainable and valued cloud offering is about more than just price. Can Eduserv, for example, compete with Rackspace’s self-proclaimed “fanatical support” without increasing costs? Can it keep up in the race to continually release new features, as Amazon is notable for doing? And perhaps most importantly, can this new education cloud become big enough to sustain its current prices or to respond to future price competition in the industry?
Finally, as the UMF Cloud Pilot leaves its beta phase and becomes a key piece of IT infrastructure upon which universities are expected to rely, it must surely begin to host applications that store and process commercially or personally sensitive data. Will Eduserv be able to demonstrate the practical security expertise that will be required? For now, the project remains an experiment, and answers to some of these questions will only be gained through trial and error. For Eduserv, the best outcome is an ongoing contract to offer this as a service to the community.