Interview with Parabon Computation CEO by Sander Olson

Here is an interview with Dr. Steven Armentrout, the CEO of Parabon Computation. Parabon has a unique approach to cloud computing – they do not own or operate any data centers. Rather, they contract out with Universities and other organizations to utilize unused compute cycles from large groups of idle computers. Parabon then leases most of this unused compute power to its customers. A portion of the unused compute power is allocated to Parabon Nanolabs, which designs drugs.

Parabon in Recent New
Parabon NanoLabs Founding Scientist, Dr. Christopher Dwyer receives the Presidential Early Career Award for Scientists and Engineers from the White House

Dr. Dwyer co-founded Parabon NanoLabs for its unique combination of DNA nanotechnology fabrication and grid computing sequence optimization that has culminated in the development of proprietary technology for precisely directing the self-assembly of designer macromolecules. Dr. Dwyer has a unique combination of wet-lab and bit-lab experience, and is a pioneer in the merged disciplines of DNA nanotechnology and computer science. He has conducted extensive research using DNA as scaffolding to support sensors that are programmed to target specific devices — for use in cancer therapeutics, bioweapons defense, and rapid readouts of DNA.

New anticancer molecules engineered to self-assemble using synthetic DNA

The grant will be used to demonstrate the viability of a new class of anticancer molecules that are engineered to automatically self-assemble from interlocking strands of synthetic DNA. It was a combination of innovations — DNA nanotechnology fabrication and grid computing sequence optimization — that led to Parabon NanoLabs’ award.

Unlike other therapeutics, Parabon’s compounds are deliberately engineered to solve specific therapeutic goals using an approach that effectively replaces the current paradigm of “drug discovery” with that of “drug design.” By affixing molecular subcomponents (e.g., antibodies, pharmaceuticals and enzymes) to strands of DNA that are pre-sequenced to attach to one another to form composite constructs, Parabon NanoLabs researchers produce therapeutics that are able to precisely target and destroy individual cancer cells, without damaging surrounding healthy tissue.

Question 1: Your company, Parabon Computation, has a unique approach to cloud computing.

Answer: Whether you call it high-performance cloud computing (HPCC) or grid computing, as we have for the past 10 years, Parabon offers the only brokered computation service in the world. We enter contracts with Universities and businesses to acquire unused compute capacity. We then aggregate this surplus capacity and make it available as an online supercomputing service, called the Parabon Computation Grid. We offer the capability to scale a computational problem across thousands of machines, and to do so within seconds. (www.parabon.com/ComputationGrid)

: How do you quantify the offering?
Answer: Our metric for measuring computer capacity is based on the cap, which is the capacity or speed an average computer. A cap-hour is the amount of work that could be performed by such a computer in an hour. So if a client has a job that requires 5,000 computers to work for two hours, he would reserve 10 kilo-cap hours. Using our system, a client can rapidly and inexpensively acquire as much supercomputing power as they need, for as long as they need it.

Question 3: How much computer capacity can a client reserve? At what point does your system become overwhelmed?
Answer: We can scale almost arbitrarily. If a client needs several hundred teraflops, we can provide that. If a compute job required computing power for weeks or months, we can provide that as well, although we might require some time to ramp up the capacity base to accommodate them, depending on the amount of capacity requested.

Question 4: Amazon charges 10 to 80 cents per hour for each CPU rented. How do Parabon’s prices compare?
Answer: We are in the range of 10 to 30 cents per cap-hour. The advantage of our system is that we aren’t just giving you a slice of some server’s capacity; rather, each server/workstation/PC is entirely dedicated to a task when it is calculating on it. But I should note that we don’t consider ourselves competitors to traditional cloud providers such as Amazon. The clients who are using our Frontier Grid Platform service routinely require hundreds or thousands of computers to accomplish their tasks, and it is difficult for most standard cloud providers to supply that level of capacity at a moment’s notice.

Question 5: What proportion of compute power on the web is wasted? How much potential is there to utilize unused compute cycles?
Answer: 80-95% of the capacity of most computers currently goes to waste. Parabon’s model therefore has an enormous amount of headroom. Demand for computing power will rapidly grow, and organizations will naturally try to find the lowest cost computer power available. For the overwhelming majority of high-performance computing tasks, we offer the most cost-effective and convenient solution.

Question 6. Parabon has recently created a new company called Parabon NanoLabs. What is the focus of this spinoff?
Answer: Parabon NanoLabs (PNL) develops and licenses proprietary macromolecules built from DNA-based nanostructures. PNL has designed molecules for cancer therapeutics, nanoscale sensors, and DNA biometrics using a CAD/CAM program called the inSēquio™ Sequence Design Studio. This application allows us to design a particular structure in software such that, by leveraging the self-assembly capabilities of DNA, we can create trillions of copies in the lab simultaneously. Designing these macromolecules is so complex that it could not be done without high-performance computing. So in this case, NanoLabs is a major consumer of the computing power provided by Parabon Computation. (www.parabon-nanolabs.com)

Question 7: Does Parabon have any plans to utilize GPU computing in the future?
Answer: Absolutely, we are very excited about the prospects for GPU computing. GPUs can actually be used for grid computing today, but we have on our roadmap an extension to our software development kit that will allow even easier access to the GPU from a grid job. This software should become available in 2010.

Question 8: How many users are renting grid computing power on a regular basis?
Answer: There are currently hundreds of users, and we expect there will soon be thousands. When Parabon began operation in 1999 there were far fewer users, but demand has grown steadily. Initially this paradigm was controversial, but it has clearly proven itself and people now understand the concept of “computation on demand.”

Question 9: One of your services actually simulates Distributed Denial of Service (DDoS) Attacks. Is there any danger that this could be used by criminal elements?
Answer: One of the services that we provide is called Blitz. This service allows organizations to test their vulnerability to DDoS attacks and therefore improve their chances of successfully defending against such. Since we alone have the application and credentials to run these DDoS operations, there is little to no risk that an actual DDoS attack could be launched using Blitz. On the contrary, Blitz allows Government agencies and corporations to strengthen and improve their defenses to DDoS attacks.

Question 10: Parabon has a capacity market. How does this work?
Answer: We seek out organizations that have large numbers of unused computers that are routinely idle during certain periods. Many universities, for instance, have large numbers of PCs and servers that are idle at night. We contract out with the organization to use a set number of computers for set periods. Currently we are only accepting bids for 50 caps of computing power or more, but we will eventually allow bids for individual computers.

Question 11: But many institutions are more interested in effectively utilizing their own surplus computer power than renting it from someone else.
Answer: Yes, and that is why we developed software to allow corporations to utilize the surplus capacity of their own computing infrastructure. Our Frontier Enterprise software can be up and running within 15 minutes; it never interferes with normal operations, and allows corporations to effectively utilize every computer they own in a customizable, private grid. Given that most organizations waste the vast majority of their computing power, this software is very powerful, and an easy solution for companies interested in increasing their high-performance computing (HPC) capability using their own IT infrastructure. (www.parabon.com/EnterpriseGrid)

Question 12: To what extent is cloud computing hampered by a lack of standards?
Answer: I don’t think that it is. If you look at the numbers, cloud computing and grid computing are rapidly being adopted, perhaps because they represent a paradigm shift that is long overdue. Both are growing exponentially and should continue to do so.

Question 13: So this supercomputing power can be directly accessed through almost any computing device, including cellphones?
Answer: Yes, the Frontier service can be accessed through almost any computing device. For example, a cell phone that has a browser can be used to drive a web-based Frontier application, providing it direct access to hundreds of teraflops of compute power.

Question 14: How much growth do you expect in the cloud computing field in the next decade?

Answer: I believe that we are on the cusp of a major transition away from traditional data centers and towards cloud and grid models. I expect to see a 20% compounded growth rate in these fields. In 2019, individuals will be renting computer power routinely, and the entire process will be almost transparent. How many machines and what exact resources are being used will be invisible to the user. They will simply specify the computational task they want completed and when they want it done, and the application will tell them how much it will cost. Accessing spare computer power will be no harder than getting electricity from a wall-socket today.