Cloud Computing – The Ultimate Wave (Part 1 of 3)
By Brian Shellabarger
This is part 1 in a 3-part series
Cloud Computing – The Ultimate Wave
The birth of giant waves
In July 2001, an entrepreneur named Bill Sharp extended the ultimate surfing challenge. The surf-wear company Billabong would offer a $500,000 prize to anyone who successfully rode a 100-foot wave. As you might expect, the idea of surfing a 100-foot wave became the overnight holy grail of tow-surfing.
Such enormous waves are exceptionally rare – created by a huge blast of ocean energy combined with immense winds which settle into a swell and collide with an underwater obstacle. This causes a huge blast of upward energy. If a surfer is lucky enough to find such a wave, most will tell you that the only thing more dangerous than riding such a monster is figuring out how to end the ride without drowning. The Bill Sharp challenge has yet to be won. To date, the largest wave ridden was 90-feet, in November of 2011, by professional surfer Garrett McNamara.
Small waves are a natural part of the evolution of industry. But the best and the brightest often confuse hype with the formation of giant waves. Enormous waves, however, are a once in a lifetime occurrence – and the ability to recognize them and successfully navigate the turbulence created by the incredible amount of energy is something few have accomplished.
I believe that the formation of an epic wave is under way. And, while many recognize that it’s coming, few understand just how enormous the impact really will be. Three incredible forces are on the verge of a simultaneous collision, which will cause cloud computing to become the ultimate wave of this generation.
Force #1: Highly Efficient Data Centers
Born more than a decade ago, the idea of buying your compute power as an outsourced commodity is nothing new to most folks who work in the technology space. Companies like Rackspace, Softlayer, Amazon, and even GoDaddy have made a nice living by selling compute as a service. However, until very recently, the value proposition was limited to net-new compute workloads as the cost of migrating existing workloads to ‘the cloud’ often exceeded the potential ROI.
Recently, however, the math has been flipped upside down. Giant data-centers are now achieving incredibly attractive economies of scale by consolidating workloads as tightly as possible. Modern applications written specifically for cloud-centric deployments are taking further advantage of these efficiencies by dynamically scaling their infrastructures up or down to meet the point-in-time demand. Where servers would previously sit idle 30 to 70 percent of time, cloud-optimized servers are often running at 90 percent capacity nearly 100 percent of the time. Modern virtualization, storage, provisioning, and automation software also means far fewer people are needed to manage the infrastructure. In fact, large-scale data center operators, such as Microsoft, are finding that, using these tools, they can operate tens of thousands of servers with the same number of people needed to operate just a few hundred. According to Cisco, a data center operating 1,000 servers is roughly 50% cheaper, per server, than a data center operating with less than 100 servers, and the numbers get even better as you add more. Bigger scale equals bigger savings.
Making this movement even more attractive is the idea that many cloud companies are offering not just to host your workloads, but manage them as well. Imagine being about to point to your most complex and laborious applications that you currently operate (Exchange, SAP, your PBX, etc.) and being able to move it into an environment where you pay only for what you use. Your costs become easily predictable, your service level improves, and you’re always running the latest and greatest. But perhaps even more compelling is that you can shift the burden of success to the vendor. If it doesn’t work, you don’t pay.
Cisco predicts that by 2014, more than 50% of all workloads will be processed in the cloud. Considering you can now by a cloud-optimized and fully managed compute instance for less than 1.5 cents per hour, the ROI is pretty easy to justify. Most businesses can’t even buy the power needed to run a server for that price, let alone the entire compute instance. I believe Cisco is under-predicting the attractiveness of the migration to the cloud, and the real number will be much higher 2 years from now.