Visit additional Tabor Communication Publications
January 15, 2009
Reston, Va.-based Parabon Computation started out of CEO Steve Armentrout's frustration at the limited amount of HPC resources available to him as a researcher. In 1999, the same year that SETI@Home was released to the public and everyone was talking about grid computing, Armentrout started Parabon in an effort to provide the infrastructure to utilize previously idle computer cycles. Parabon -- along with companies of that era such as United Devices (now Univa UD), Entropia, Process Tree, and others -- offered users an opportunity to contribute their cycles to cancer research (an effort it still supports), and enticed volunteers to donate their idle cycles with monthly sweepstakes.
But that was just to get things started. Parabon soon created a business model from its cycle scavenging software, the Frontier Grid Platform. Today the company offers customers the ability to purchase CPU time on a network of servers ("brokered computation") or, for larger enterprises with significant existing IT deployments, the ability to install Frontier in-house to take advantage of excess capacity in its own infrastructure. What kinds of work do Parabon's users do with the resources its software rescues from disuse? "Frontier's users are in energy, finance, DoD, nanotechnology, and other areas," says Armentrout.
The DoD is one of Parabon's large recent wins. In December of 2008, Parabon announced that the Defense Contracting Management Association had inked a deal to deploy its Frontier Grid Platform on 10,000 computers within the DoD. DCMA plans to use Parabon Crush, the company's distributed statistical modeling application, to run financial forecasting models on its over 300,000 active contracts.
As you would expect, Frontier is heavy on the ability to farm out a slew of embarrassingly-parallel jobs -- like parameter space searches, for example -- onto idle resources. Frontier provides an API for developers who want to tightly integrate their application with Parabon's service, and also provides a tool for users to use unmodified applications and farm them out over hundreds of servers. In recent years Parabon has also added the capability to schedule and manage more traditional (i.e., tightly coupled, MPI-sporting) scientific applications as well.
But if you aren't the owner of 100,000 idle desktops over in the Provo Office, and you are a Parabon Computation on Demand customer, where is your job actually running? "Parabon leases idle cycles from the unused resources on university campuses and businesses," says Armentrout. The company owns no hardware, and CEO Armentrout has no plans for that to ever change. Most of its brokered cycles are homed in the US, as are the bulk of Parabon's customers, but Armentrout says the company has a growing presence outside the US as well.
The "get this work done for me somewhere, I don't care where" mindset in Frontier has been central to the company's development from the beginning, and users of Frontier-managed resources don't manage servers or connections. Instead, users manage parcels of work to be completed and job submission, server failure, coordination, and communication are all managed for the user by the software. Parabon says that Frontier is platform agnostic, supporting Linux, Windows, and Mac OS X.
But how is all this different from, say, what Amazon is doing? Armentrout draws a distinction between cloud computing offerings, like Amazon's EC2, and Parabon's Grid offerings. "EC2 offers a lot of automation and infrastructure for customers to very rapidly deploy a virtual machine, say for e-Commerce, and grow or shrink that capability to match demand. But it's not principally a computation service," says Armentrout. "Our own benchmarks show that a dime spent buying a computation hour from Parabon gets users about three times the capability of that same dime spent in Amazon's cloud."
He explains that this is primarily due to the fact that Amazon may deploy several VMs on each physical server, to ensure that it is getting maximum use of the resources. Frontier users get all of the server they are running on at the time. Frontier and EC2 users alike must provide the application, any libraries, and relevant data files, however the submission process to the respective services highlights their differences. Whereas Amazon users provision VMs and then customize them one-at-a-time (which suits the Web application use case), Frontier has an API that accepts programmatic submission of thousands of computational work requests, which it then automatically executes across available resources, rescheduling work from machine to machine, if it must, to guarantee timely delivery of results.
Armentrout also points out that Parabon's "no hardware" model has enabled it to be successful while other pay-for-computation services, such as Sun's Network.com (which recently closed to new users), have struggled, or failed outright. Although the lack of data available for this privately-held company makes it hard to assess just how successful it has been.
Parabon has been in the news this week as NASA announced the award of a Phase 1 Small Business Innovative Research Grant (SBIR) for feasibility testing of an idea that Parabon calls "Componentized Models as a Service." The idea leverages previous investments that NASA has made to transform the computational building blocks of its scientific models -- a particular oceanographic model for example, or a specific atmospheric discretization -- into components that can be wired together to create a custom application that addresses a particular need. Parabon's task is now to create a Web portal that allows users to compose applications from these components, dynamically compile them into a new application, and then distribute the computation to NASA's compute infrastructure using the Frontier Grid Platform. This is an interesting twist on the often tried, but frustratingly seldom successful grid portal model which simply makes a new interface for existing static applications.
Parabon is partnering with Goddard on this project, which Armentrout says should take about six months. If this trial proves successful, the company could enter into further agreements to expand and enhance the idea for NASA. Armentrout describes Frontier as "a service approach to HPC without the expense of hardware." With this latest announcement the company is trying to parlay its decade of experience in "getting the work done" into a new generation of solutions that help users specify what work is to be done, and how (computationally) to get it done. This kind of work is critical to lowering the usability barriers that IDC and others consistently identify as fundamental obstacles to the widespread adoption of HPC in new and nontraditional areas of computing.
Jun 19, 2013 |
Supercomputer architectures have evolved considerably over the last 20 years, particularly in the number of processors that are linked together. One aspect of HPC architecture that hasn't changed is the MPI programming model.
Jun 18, 2013 |
The world's largest supercomputers, like Tianhe-2, are great at traditional, compute-intensive HPC workloads, such as simulating atomic decay or modeling tornados. But data-intensive applications--such as mining big data sets for connections--is a different sort of workload, and runs best on a different sort of computer.
Jun 18, 2013 |
Researchers are finding innovative uses for Gordon, the 285 teraflop supercomputer housed at the San Diego Supercomputer Center (SDSC) that has a unique Flash-based storage system. Since going online, researchers have put the incredibly fast I/O to use on a wide variety of workloads, ranging from chemistry to political science.
Jun 17, 2013 |
The advent of low-power mobile processors and cloud delivery models is changing the economics of computing. But just as an economy car is good at different things than a full size truck, an HPC workload still has certain computing demands that neither the fastest smartphone nor the most elastic cloud cluster can fulfill.
Jun 14, 2013 |
For all the progress we've made in IT over the last 50 years, there's one area of life that has steadfastly eluded the grasp of computers: understanding human language. Now, researchers at the Texas Advanced Computing Center (TACC) are utilizing a Hadoop cluster on its Longhorn supercomputer to move the state of the art of language processing a little bit further.
05/10/2013 | Cleversafe, Cray, DDN, NetApp, & Panasas | From Wall Street to Hollywood, drug discovery to homeland security, companies and organizations of all sizes and stripes are coming face to face with the challenges – and opportunities – afforded by Big Data. Before anyone can utilize these extraordinary data repositories, however, they must first harness and manage their data stores, and do so utilizing technologies that underscore affordability, security, and scalability.
04/15/2013 | Bull | “50% of HPC users say their largest jobs scale to 120 cores or less.” How about yours? Are your codes ready to take advantage of today’s and tomorrow’s ultra-parallel HPC systems? Download this White Paper by Analysts Intersect360 Research to see what Bull and Intel’s Center for Excellence in Parallel Programming can do for your codes.
Join HPCwire Editor Nicole Hemsoth and Dr. David Bader from Georgia Tech as they take center stage on opening night at Atlanta's first Big Data Kick Off Week, filmed in front of a live audience. Nicole and David look at the evolution of HPC, today's big data challenges, discuss real world solutions, and reveal their predictions. Exactly what does the future holds for HPC?
Join our webinar to learn how IT managers can migrate to a more resilient, flexible and scalable solution that grows with the data center. Mellanox VMS is future-proof, efficient and brings significant CAPEX and OPEX savings. The VMS is available today.