Jason Stowe, CEO, Cycle Computing
Innovate or die. How many times have we heard this mantra in our careers? The difference is that today’s modeling and simulation tools will move CIOs to the forefront of discovery and innovation.
“Cycle Computing software dynamically provisions dozens to hundreds of thousands of cores to create a cloud cluster computer for those simulation, modeling, and analysis jobs today”
While “faster time to market” has long been a corporate mantra, a confluence of several critical technologies, are driving a new reality. Cloud-driven analytics and simulations are driving significant cost and time out of product development cycles. Manufacturers are designing hard drives in half the time of two years ago and drug companies are accelerating discovery processes fivefold.
And while the CEO’s vision of bringing better products to market faster may make the R&D department tremble - insightful CIO’s are seizing the opportunity to demonstrate that IT can have a direct and quantifiable impact on line of business success as the engine of innovation for the next century will be analytical and simulation. It will start with the troves of data from customers, instruments and databases that will be collected, organized, stored and analyzed for insights for better understanding of how results were generated.
But the real business value will be created when those insights are transformed in foresight through analytical models. Some of the most powerful tools researchers and engineers use to speed their development efforts are tied directly to analysis and simulation - or in today’s hype vernacular, Big Data and Big Compute. While the tools to perform both have been available for years, cloud has opened new pathways to virtually limitless compute and storage resources, in highly cost-effective models.
Until recently, simulation and analysis at this larger scale, was only available to those large organizations with enough resources to purchase a supercomputer. Over the past decade, access has increased with the adoption of cluster computers. But the greater availability of clusters to researchers only led to greater headaches for those tasked with administering, housing, powering and cooling them. The missing key to greater adoption has been large scale access to more computing, which the cloud enables.
In the case of big compute, innovative companies are in fact leveraging the cloud to do more simulation and analysis, which is leading to better answers, faster. And those better answers lead to better decisions, and better products, all faster. The early adopters of big compute have already proven this model. So, if your business hasn’t implemented a big compute strategy yet, while it’s not quite panic time, it is time to get moving.
Real examples. Real impact.
Cycle Computing software dynamically provisions dozens to hundreds of thousands of cores to create a cloud cluster computer for those simulation, modeling, and analysis jobs today. Call it on-demand computing, where scientists and engineers ask the questions they need to ask, rather than the question they think their existing infrastructure can handle. The net result is that when the right question can be asked, the answer is far more valuable. That is the heart of innovation and the aspiration of every R&D department.
One innovative company, Novartis Institutes for Biomedical Research presented at a recent AWS re:Invent how their researchers are leveraging the cloud in their race to find drugs that will defeat cancer.
The problem: a research project targeting a specific protein associated with some cancers required 50,000 CPU cores to run optimally. Novartis had 7,800 available.
The solution: Novartis partnered with Cycle Computing and AWS to complete 39 computing years of science achieved in 11 hours, and an estimated $44 million in computing infrastructure accessed for under $5,000.
This effort, initially a test proof of concept, actually led to three compounds that scored high enough to create the promise of becoming part of a new drug to fight cancer. This work, quite simply would not have been possible two years ago and will be standard practice going forward.
One of the keys to success of the Novartis example is the speed with which the Novartis IT organization was able to respond to the researcher. Being responsive to user demand is vitally important in this equation, and it requires a great deal of agility on the part of businesses. Cloud cluster computing now allows our customers to shorten the time to result for the user while being able to know exactly how long it will take and how much it will take before the computation begins. Regardless of where one feels we are on the cloud hype cycle, these results are real, quantifiable, repeatable and truly disruptive over the historical methods.
Access to computing power is only useful when simulation and prediction software is extremely accurate. It’s always cheaper to simulate than prototype. This removes most false starts and bad ideas and, overall, it accelerates business. Today’s big compute tools on cloud allow subscribers to the notion of fail fast, to fail faster than ever - and ultimately succeed.
Another example, from Compendia Bioscience, was presented at re:Invent, where the company’s Gene Fusion Detection Science project has been able to analyze more than 10,000 DNA and RNA samples with the goal of providing better data to fight cancer. In this process, Compendia Bioscience was able to analyze more than 4,000 tumor samples, 19 cancer types, process 32.2 TB of data, and accomplish 15.6 computing years of analysis in just months.
Business Units Demand Cloud Cluster Computing
As costs and complexity continue to drop, demand has increased for big compute at the business unit level as well. Typically one business unit at a company achieves the agility and cost-savings, and other units soon follow. And once the CIO validated and institutionalized the process it soon becomes a pathway to fulfill the unmet needs of other business lines they serve.
Rapid Adoption Paths
Rarely discussed but always appreciated, an additional benefit of cloud based big compute is just how quickly ideas can move from concept to production. A pilot program can take anywhere from one week to a month to implement. Next, users have the ability to manually conduct burst to cloud, and cloud-only runs of specific workloads, leveraging internal and external resources. As the experience proves out a high ROI, with a new business agility to meet IT needs, they typically move towards a more automated model, implementing one-click cluster containers that can be spun up by individual researchers or engineers on the fly.
The Hero CIO - Deliver the Tools that Speed Time to Market
Analytics and simulation to CIOs in 2015 are much like mobile computing in 2000. Some saw that mobile devices would not only transform how users interacted with each other, but how IT would need to transform to serve users and customers. Some CIOs guided their companies through it; some started completely new companies based upon it - while others failed miserably. Analytics and simulations have started on a similar trajectory but on an even more accelerated path. If industry follows an arc similar to mobility we are in for a wild ride of innovation across all facets of commerce and society. And CIOs will have been right in the middle of it all.