The Internet as we know it is imperfect — that’s why the GENI ultra-high-speed test bed is helping researchers develop technologies that will allow people to do things that are usually impossible on today’s network.
It was sometime in 1998 that most people suddenly became familiar with the Internet. In 1997, people called it “that Web thing,” and they would confuse email addresses and URLs — generally no one paid the whole affair too much mind. The transition that people talk about today — between spending hours trying to remember some fact to just looking it up online — happened in less than a year.
Society is rapidly approaching an era in which technology is no longer outmatched by even the most fantastic trappings of human imagination. It’s miracle enough that instantaneous remote communication via the telephone is commonplace, but each day researchers are pushing the boundaries of what people think is possible. And when a good idea is found, it can enter society and begin making subtle and widespread changes in behavior, landscape and culture in a matter of months.
The Internet is the foundation of 21st-century technology, but the Catch-22 of developing new Internet technologies is that scientists are often constrained by the very environment they are attempting to innovate. A test bed for researching new networking ideas was needed, and that’s why the Global Environment for Network Innovations (GENI) was created by the National Science Foundation in 2007 and has since received $80 million in federal investment.
GENI is a network of more than 50 sites in more than 30 countries that allows its 3,700 member researchers to test their ideas in a low-latency, high-bandwidth digital environment unconstrained by the foibles of the regular Internet. Before cutting-edge networking technology can enter the real world, it needs to be tested on a broad scale, and GENI is where that testing happens.
In March, GENI partnered with US Ignite, a nonprofit dedicated to developing next-generation Internet apps, to lead a three-day event that showed off what they could do. Event coordinators demonstrated to an audience in real time how a low-latency fiber network could enable functionality that public safety officials can only dream about using in the field today.
In one demonstration, a connected vehicle simulated a crash, which triggered an automatic notification to a police station. Instantly police knew where the crash was and which vehicle was involved. A video-equipped drone was immediately deployed to the site to collect information, like whether there was fire, hazardous chemicals or other threats that might warrant the use of special equipment. The crash was fake, but everything else was real. The audience watched everything happen via a video feed delivered at the speed of light, and though today’s infrastructure and policy can’t support such an implementation in the real world, the technology for this type of application is ready when the people are.
Today’s Internet is constrained mainly by three things. The first is bandwidth. There’s not enough uniform connectivity across the Internet to ensure that content can be delivered at the quality and speed necessary for sophisticated new applications to perform effectively. The second is latency. Even relatively low latencies like 100 milliseconds are far too slow for real-time applications that require high precision. The third is security. Security issues will never be thwarted completely, but today’s Internet can’t be used for certain applications in good conscience — remote surgery, for instance. But GENI technologies, said Glenn Ricart, CTO of US Ignite, allow people to do things that are usually impossible on today’s Internet.
Software-defined networks (SDN) and other software-defined technologies are a big part of what makes GENI so innovative. SDN is changing how private companies ply their networks thanks in part to GENI’s demonstration of how powerful these technologies are. Companies like AT&T, Cisco, Comcast, Juniper Networks, Amazon and Hewlett-Packard support US Ignite and stand to gain big from the advances demonstrated by GENI’s network. One unique component of SDN is that it allows Internet architecture to be divided into slices or channels.
“The basic idea of today’s Internet is that both the control signal and the data signal are being run over the same Internet,” Ricart said. “SDN is the notion that you separate the data and the control into separate channels, so that way when you want to have a connection to a health-care provider, the SDN checks [that the user has adequate clearance to do this] and then it creates what’s called a flow, keeping it separate from other traffic.”
These software-defined slices of Internet provide several benefits, security among them. If the Internet is to be used regularly for things like dissemination of electronic medical records or the aforementioned public safety application, segregating functionality into different slices provides an additional layer of security beyond what’s possible today. It would also let Internet service providers sell users varying levels of service, like access to a health-care channel or a financial services channel, Ricart suggested. While such commercial arrangements raise net neutrality issues, the concept could add functionality to the Internet that simply isn’t available today.
“For example, if someone had a home dialysis unit and wanted a secure medical channel, they could have that,” Ricart said. “That might be paid for by health-care insurance, and that gives a low-latency, high-quality channel with highly reliable monitoring of a home dialysis machine or home infusion pump for cancer drugs. We hope to have more health care delivered at home, so if a doctor needs to go and see if a diabetic is healing properly, being able to get 4 Kbps video upstream from the person’s home to the hospital means that we can save an ambulance run to take that person to the hospital.”
SDN also mitigates bandwidth limitations and reduces latency by distributing data across a network. In another demonstration by US Ignite, researchers in five cities — Potsdam, Germany; Brussels, Belgium; Victoria, British Columbia; Washington, D.C.; and Tokyo — collaborated on a huge database project in real time thanks to minimized latency and distributed data. Rather than transferring and updating the project’s huge data sets with each edit, small signals were instead sent between the collaborators, and their local applications reflected the changes being made across seas. To accomplish this low level of latency, copies of the data were distributed physically close to the user in each country, rather than each user drawing from a central database.
On today’s Internet, conversely, thousands of users across the nation might access the same video streaming service at the same time and draw the same information from the same far-flung data center, pulling gigabits of redundant data across thousands of miles — an arrangement less efficient than if the data were distributed nationwide and accessed from across shorter distances.
GENI’s products, particularly an SDN open standard called OpenFlow, are making their way onto the regular Internet. GENI’s first project director, Chip Elliott, said it’s just a matter of time before the software behind the apps people see on their screens begins using networks in smarter, more innovative ways.
“I think this is really going to change the way the whole Internet is built and operates pretty soon,” Elliott said. “I’m not saying GENI will change it, because GENI is a research project, but a lot of the technology we pioneered is moving into the commercial space very quickly. OpenFlow has completely taken the data center world over. People are enthusiastic about it, and it’s being applied in long distance as well. You’re starting to see the Department of Energy, the Department of Defense and various other groups who are beginning to think this could be handy for the way they do business.”
Elliott recognized the potential conflict between a preservation of net neutrality and the concept of Internet channels. Millions signed a petition last year asking the FCC to preserve net neutrality, and though it’s expected that the reclassification of Internet service as a Title II utility under the Communications Act of 1934 will preserve net neutrality in the short term, a new technology like SDN, no matter how useful, would inevitably raise the public’s ire if it upended that hard-fought victory.
“The word ‘channel’ itself is a red flag, but this type of technology really does give ways to either do net neutrality or not do net neutrality. It’s a powerful tool,” said Elliott. “My own view is net neutrality is basically a policy decision. People already have enough knobs to control whether they favor something over something else, and so ultimately it has to be resolved at a policy level. And the tech is growing fast now. There are versions of the equipment from Cisco, IBM, HP, Dell, Sienna. People are starting to buy it and just plug it in. It really is snowballing quickly at this moment.”
One of the most exciting applications of GENI technology is the potential to speed the transformation of industries like medicine. GENI Project Director Mark Berman noted that researchers at the University of Chicago and elsewhere are using GENI to create a bioinformatics exchange to share huge repositories of genomic data for cancer research and to advance personalized medicine that lets clinicians tailor care to individuals’ genes and cellular biologies. President Obama allocated $215 million in his 2016 federal budget toward building a personalized medicine infrastructure, mainly through projects at the National Institutes of Health and National Cancer Institute.
“We’re getting to the point where the data that my doctor might need to customize my personal treatments based on my genomic data could be available at a price that makes sense,” Berman said. “However, that doesn’t solve the problem because the problem remains that even if my doctor has that data, it’s not clear she can do anything with it. Doctors are going to want to compare my genome against databases of many thousands or tens of thousands of other patients. So we need to couple some very powerful analytics with some powerful networks.”
In the coming decades, discoveries in the human genome and their applications in personalized medicine will make today’s medicine look primitive. The cancer drug Avastin, for instance, is often used in treating non-squamous cell carcinoma, but differentiating between squamous cell and non-squamous cell can be difficult even for experienced pathologists.
“You want to get this diagnosis right, and you can do it quite accurately if you compare the genomes,” Berman explained. “This is a good example of personalized medicine, but you can’t compare the genomes until you can move the data around and get the analytics in place. And that’s the kind of thing you can do with GENI.”
When people think of improving the Internet, they think it’s all about increasing bandwidth, but there’s much more than that, said Berman. It’s also about detailed data flow control, minimizing latency and intelligent software that can handle fast-moving mobile devices. “In the case of a moving ambulance, you’re moving in and out of coverage, you may want to change the resolution of your video or prioritize patients’ vital signs data over a video stream or something like that, and these are things that are not quite impossible but really hard to do with existing wireless networks,” Berman said.
One research project led by Rutgers University, called MobilityFirst, integrates with GENI to issue notifications without relying on a traditional centralized host server. Projects like this show that developing the technology to support innovative ideas is no longer the highest hurdle, because the technology itself is basically ready. The barrier is that cities are missing the roadway network infrastructure as well as the strong governance and policy to support those technologies.
“I think we’re going to change how people think about the Internet,” Berman said. “What GENI and some of our projects are going to do is make that control aspect accessible to people who are building important applications, and that will allow them to do a lot of things they can’t do today.”
Students from George Washington University pushed the bounds of today’s software in a $10,000 GENI software contest sponsored by Cisco and launched in conjunction with the school’s 22nd GENI Engineering Conference. Winners developed tools for real-time SDN monitoring, load balancing and dynamic topology modification, and packet mapping. GENI technology is exciting because it opens new areas of research and gives developers greater control over what their apps do, said Donald DuRousseau, the university’s director of Research Technology Services.
“It really gives us the ability to operate networks at scale because that’s what GENI is about,” he said. “It gives us a chance to run things across the country, not just a simulation in a lab.”
One barrier between the realm of research and the mainstream is the capability for automatic configuration. Complex technologies can be sold to the public today thanks to software conventions and standardized protocols that don’t require the user to have any knowledge of the product’s mechanics. That’s why almost anyone can plug in a wireless router and get a home network running without needing to configure ports and set IP addresses. GENI is approaching the point where its underlying technologies will be ready for public consumption, so the market potential is enormous, DuRousseau said.
“It’s a multibillion-dollar potential market space, just in the GENI domain itself,” he said. “At our conferences, we’ve gone from letting everyone know that this technology exists and getting people interested in it, to getting it into more people’s hands to actually use it and develop. I think it’s reached a pinnacle.”
In one US Ignite conference demonstration, a 5 Kbps camera mounted on a microscope at the University of Southern California relayed video to students in Chattanooga, Tenn., who could view and control the microscope’s stage in real time. The demonstration showed how a reduction from a typical latency of about 100 milliseconds to an SDN latency of about 35 milliseconds made the stage’s movement appear to be instantaneous and made the application much easier for users. But setting up the demonstration wasn’t easy. Organizers reported spending an hour on a conference call with five network managers who each needed to manually adjust their SDN settings so everything would work properly. Automating the process of establishing such a low-latency flow is work that researchers will focus on in the next three to four years and could bridge the gap to the mainstream.
GENI technology like OpenFlow was even adopted by the very network that hosts GENI, an innovation test bed called Internet2. “It’s a virtuous circle,” said Rob Vietzke, vice president of network services for Internet2. “In some regard, we were created to support things like GENI, and they are created to incubate technologies that we would adopt and bring to deployment.”
Internet2 moves more than 60 petabytes of data across its network each month, with parallel purposes to GENI, as the network supports things like genomic research, astrophysics and remote learning. How long until the premarket technologies researched under Internet2 and GENI enter the mainstream depends on the technology, Vietzke said, but it can happen very quickly.
“What we want to do is let people with fairly low resistance try new ideas and adopt them, and fail as well as succeed,” he said. “For some things, like what happens with OpenFlow and SDN, you saw this massive change in the way industry was thinking about developing networks over a very short period of time. You can’t imagine that every rural health clinic is going to adopt ultra-high-definition microscopes in the short term, but you can imagine that major research hospitals in each part of the country will start to do that within 18 months or so.”
Vietzke observed overlap between the work he does in cutting-edge networking technology and his experience working for state government.
“A lot of [IT managers] deal in constrained environments where they are constantly fighting with just getting a little more capacity, capability to solve urgent problems,” Vietzke said. “What’s happening in US Ignite, GENI and Internet2 is they are helping to create a platform where local innovators find ways to do innovation without having to make a major investment.”