Cloud war collateral: What the rise of AWS, Azure has meant for data centers
FYI, this story is more than a year old
When Henry Ford introduced the Model T in the fall of 1908, he likely didn’t comprehend the full scope of events he would set in motion.
Come 1914, and Ford’s production line had reduced assembly times from 12 hours to less than two and a half hours, slashed the going price of an automobile, and redefined the working wage of factory employees, ultimately putting more than 15 million Model T’s on the roadOpens in a new window and igniting the entire automotive industry in the years to come.
Competition often leads to innovation and progress for other industry players. One modern equivalent of this can be seen in the rise of public and private cloud providers like Amazon and Microsoft.
AWS’ sales numbers recently topped $12 billion, up nearly 55 percent from the same period last yearOpens in a new window. Meanwhile, Microsoft continues to push ahead and is projected to reach $20 billion in annual cloud revenue by June 2018.
As these powerhouses and others like Oracle and Google continue to see widespread adoption across industries, other players have stepped in to consume their piece of the $204 billion-dollar cloud infrastructure pie, leading to an ecosystem of cloud and data center partners that continue to push the technology envelope to expand capabilities of these offerings.
Speaking to changes in our modern data centers, here are some of the key advancements:
Real-Time Monitoring and Analytics – With connectivity impacting virtually every facet of a consumer’s daily lives from personal to professional, the need for granular understanding of how the data center is operating in real-time has never been more important.
Data and analytics empower the unsung heroes – data center managers – to make on the spot decisions and adjust workloads as needed, ensuring business as usual is maintained.
All data center equipment, with all this computing power, needs to be kept in a ready state of health to deliver their services. Each server and each network node needs to be monitored continually to ensure sustained uptime.
Greater Consideration to Traffic Needs and Prioritizing Uptime – Being critical to business workflows, data center outages commonly make headline news and can quickly account for significant losses in revenue and customer trust, even in the briefest periods of downtime.
Proactive planning for traffic spikes, outage response strategies, and secondary systems become central initiatives for data center managers, while information collected from those real-time analytic readings allow managers a greater line of sight into potential problems and identify areas where routine maintenance is necessary.
Got Power? – Data centers have moved beyond being the equipment in the ‘back room’ to large free standing facilities, where a collection of 50,000 servers is not uncommon and a few select data centers run upwards of 100,000 units to handle the unprecedented workloads.
With this scale, the ability to support the data center on the power grid can be a real consideration in terms of data center siting and maximum capacity of facility. Facilities on the scale of 10MW, 20MW are very large but realistic potential footprint at the very top end. Large HPC data centers need to be designed based on the total available power envelope.
This requires exquisite design and instrumentation to maximize performance in the most efficient manner, at the largest scale.
New Workloads and Proliferation – Just as the data centers themselves expand, the work they can perform becomes more segmented and specialized.
With the increase in scale and corresponding reduction in cost per cycle of compute, new workloads become cost effective to execute – personalized Genome sequencing is an example of something that has come into the range of a consumer service, given the reduction in compute cost.
These innovations would have been reached eventually, much in the same way that another car company would have figured out putting employees on a moving assembly line to do one job was a faster way to put a car together.
Yet the demand for new offerings from these tech giants and the massive financial opportunity in the data center market has accelerated product pipelines and driven engineers to rethink existing offerings, allowing cloud companies to truly revolutionize how we store, access, and share data.
Article by Jeff Klaus, Intel IT Peer Network