Public Cloud and 3rd Party Infrastructure Optimization Software
By: Evan Smethurst
www.techrealestatetrends.com
The Public Cloud has become a massive force over the last several years. The major Infrastructure-as-a-Service (IaaS) providers have designed their Cloud Computing platforms for the ‘public’ at large and as a result there has been significant demand. The spectrum of demand has been comprehensive: startups, mid-market, corporations and the world’s biggest enterprises. An overwhelming percentage (95%+) of mid-market and enterprise companies are leveraging Public Cloud. The industry’s dominant company, Amazon Web Services (AWS), released financials showing 69% growth and $7.8B revenue in 2015. This incredible influx of users demands a significant engineering effort, and the major IaaS players have responded, architecting near-infinite scalability within their Clouds.
Near-infinite, on-demand computing scalability has its risks for the Cloud Computing vendor. Imagine a cross-section of a ‘hyperscale’ Cloud Computing data center in which a major IaaS provider operates: racks upon racks of horizontally stacked server, storage and networking gear, symmetrically deployed to make the most efficient use of space. To comprehend the immensity of a Cloud facility, visualize a data hall inhabited by computing racks endless enough to justify Segway transportation for the facility operations staff. Enterprise computing hardware doesn’t come cheap, but the IaaS business model necessitates that compute is available during peak business times. Since the IaaS vendors have promised the marketplace fully ‘elastic’ availability of computing resources, they take the risk of investment into ‘burst capacity’: the infrastructure deployed to support peak business demand. Inherent in a ‘burst capacity’ investment is the risk that the associated infrastructure commonly sits idle, waiting for paying users to rent Virtual Machines (VMs) on an hourly basis in order to generate revenue.
The executive staff at a leading global IaaS provider finally grew tired of the intervals in which their expensive investment into ‘burst capacity’ sat idle, and decided to create a stimulus for users: the Spot Market. The computing ‘instances’ (actively ‘leased’ VMs) on the Spot Market are functionally identical to the instances leased ‘on-demand’, but carry a tantalizing discount of up to 90%. Yes, there is a catch: ‘Spot’ Instances remain available for lease on the open market, even while it might be crunching data as a part of your computing fleet. This point is worth emphasizing: a user outside your network could offer a higher price for your Spot Instance, and the IaaS provider could unceremoniously remove the computing resource from your possession for presentation to the high bidder. By original design, these business continuity risks made Spot Instances appropriate only for development, Quality Assurance, and other non-production IT workloads.
Thankfully for the Cloud computing user, some of the most ‘gamechanging’ technology developments of 2016 include ‘Intelligent Workload Management’ (IWM). IWM in the Cloud includes the automation of bidding and failover processes for Spot Instances such that they can be used in a production-ready, customer-facing computing environment. There is no doubt this recent innovation has completely changed the economics of Cloud computing. With intelligent software-enablement, your IaaS compute is now 50-80% less expensive.