Hardware in the Cloud: Less Really Is More
The cloud is doing more than simply changing the way we humans interact with data infrastructure. It is altering the very make-up of the infrastructure itself, and potentially ushering in some turbulent years for the well-established manufacturers that have come to dominate the IT industry.
The first wave of this change is upon us already as vendors seek to expand their portfolios of traditional enterprise-focused technology to new cloud-optimized systems and solutions. In the cloud, high density, multitenancy and near universal virtualization rule the roost, and the fact is that as more data loads are shifted away from traditional data centers, demand for old-style infrastructure is likely to diminish.
For the designers and manufacturers of IT technology, then, success or failure will rest largely on how well they can transition their product lines from the static, silo-based infrastructure of old to the dynamic data environments of both public and private clouds. This struggle is already playing out on the processor level with Intel, AMD and others engaged in a fierce debate over how to accommodate the future. In AMD’s case, the answer is clear: Get a jump on ARM technology while chief rival Intel seeks to repurpose x86 for the new millennium. A key issue for both platforms is whether we are seeing just a new cloud version of traditional client/server architectures, which would favor a retooled x86, or an all new cloud/mobile paradigm, which would most certainly favor ARM.
Of course, some traditional IT vendors seem to be weathering the transition better than others. According to Synergy Research Group, Cisco Systems has pulled into the lead in the cloud infrastructure equipment market, surpassing both IBM and HP. As a primarily networking company, Cisco has the advantage of having minimal exposure to the server market, which has seen a dramatic drop in activity over the past few quarters. As it stands now, Cisco is in command of about 15 percent of the cloud equipment industry.
Is it possible, though, that server manufacturers need only retool their product lines to regain momentum? Perhaps, but don’t expect servers to be as dominant in the cloud as they were in the enterprise. Cloud providers are becoming highly adept at maximizing server power and pushing resource utilization ever higher, so much so that, as the Lawrence Berkeley Lab notes, today’s fleet of approximately 3.5 million email servers could be replaced by a mere 47,000 in the cloud. Application servers could be cut from today’s 1.2 million to only 32,400 – a 97 percent decrease. This would be a tremendous benefit in terms of capital and operating budgets, but it would be near-death for the server industry.
Indeed, the market for hosted virtual servers is moving ahead at a rapid clip. ITX Design, for example, is a web hosting provider and site development firm offering a new line of dynamic cloud servers that offer broad scalability and multi-tier management capabilities without locking users into dedicated hardware configurations. In this way, organizations can be matched to an appropriate set of resources that can be reprovisioned practically on the fly as data loads and other requirements change over time. You also get instant fail-over and 24-hour service and troubleshooting without having to build an internal IT team.
Cheaper, easier and more flexible do not necessarily mean better, however, and the enterprise still has serious reservations over cloud reliability and availability. It is fair to say, though, that as reliance on the cloud for non-critical loads increases, the fear, uncertainty and doubt (FUD) will erode over time, perhaps even to the point where the cloud will become the ‘normal’ approach to IT before long.
For that to happen, the cloud will have to maintain or even surpass its current ROI proposition, and the best way to do that is to place as much data on as little hardware as possible.