08 Mar Milwaukee Institute seeks to build computational power
Milwaukee, Wis. – Private sector leaders in Milwaukee and southeastern Wisconsin are trying to bridge the gap between universities and businesses through more effective use of computing and scientific resources, and their vehicle is the new Milwaukee Institute, a non-profit organization that is building a cyber infrastructure of shared, grid-based computing.
The initial stages of the computational-power initiative has been funded by four metropolitan-Milwaukee companies, and its collaborative model is designed to lure federal grant money to sustain the effort in the long run. One of those companies is Mason Wells, a Milwaukee-based private equity firm whose executive managing director, John Byrnes, believes something significant must be done to spur innovation in the region.
“We’re trying to jump start the high-tech segment of the economy here by providing an IT-based collaborative network that will allow people to work together more efficiently,” Byrnes said. “We’re trying to give them access. In some cases, people don’t have access to computational resources they need.”
Institutionalized collaboration
The genesis of the institute dates back to the interest of the Wisconsin Technology Council and individuals like Paul Peercy, dean of the College of Engineering at the University of Wisconsin-Madison, to combine educational and research activities in southeastern Wisconsin with the business community. John Byrnes said this has not been effectively done through existing channels, and the institute’s backers have devoted time to closing that gap during the past two years.
They have organized the institute around an agenda that’s intended to support collaborative research using a shared information technology infrastructure. The local educational community has expressed an interest in sharing IT infrastructure, particularly for heavy computational science activities, but although Madison enjoys such a resource in the Condor project, southeastern Wisconsin has lacked this capability. Local research institutions are interested in developing their own research, and as they grow their respective institutions, the Milwaukee Institute intends to work with them to put in place the infrastructure necessary to support their activities.
That infrastructure will come in the form of a value-added network that the institute will create and manage. It will have a platform to help researchers manage their collaborations with colleagues inside and outside the metropolitan area, and with business research professionals who will tap into the system. According to Byrnes, those collaboration services will be augmented by computational resources, including middleware that will be designed to assist researchers in developing efficient ways of conducting their research so it can be done in a grid environment using multiple processors.
“That whole type of system is already in place in Madison,” said Byrnes, referring to Condor, a distributed computing project. “Dr. [Miron] Livny is the one who built it, and it’s being used extensively on the Madison campus. We’re trying to bring some of those ideas and capabilities to the local community here in southeast Wisconsin.”
Livny, a computer science professor at UW-Madison, works with the Condor project.
In metropolitan Milwaukee, Byrnes and Jay Bayne, executive director of the Milwaukee Institute, report positive reaction from researchers at the Medical College of Wisconsin, the University of Wisconsin-Milwaukee, Marquette University, and the Milwaukee School of Engineering.
Byrnes cited two examples – both at the Medical College of Wisconsin – of how this cyber infrastructure could be applied. The Medical College is conducting computational research into rat genomics, and genomic calculations require massive amounts of data. In addition, there are efforts underway at the Medical College to collect and analyze population data, where millions of files must be analyzed and collated.
“They have resources there now, but they are not able to share them as well as they’d like to,” he said, “so they are talking to us about developing a platform that they can tie into.”
Bayne, who also is an adjunct professor at UWM, said no determination has been made as to the lineup of projects that will be the first to run on the system. There are a number of candidates, including the computational needs of the UWM Water Institute related to hydraulics and water flow, but each institution has its own priorities and the institute will base resource-management decisions on constituents’ needs. “We’re going to eat this elephant a bite at a time,” he said.
Flight of the Condor
Condor, a distributed computing project at the UW-Madison, is an architecture designed to deliver processing power calculations that otherwise could take weeks or months to complete, and allows the execution of projects across a computer cluster. Condor incorporates about 4,000 servers on the Madison campus, allowing researchers to harness supercomputing power without needing actual supercomputer hardware.
“The idea behind Condor is to enable people to pool together their computing resources and benefit from them in a community,” Livny said.
With a Condor-like approach, that supercomputer environment is replaced by a system consisting of many smaller pieces – computers and servers – that provide a composite answer. A comparable example is a program of the Pennsylvania Infrastructure Technology Alliance (PITA), in which people volunteer their personal computers when they are idle and combine that with software that seamlessly taps into idle processors in a grid to utilize massive amounts of computer power that is not all in one machine.
Livny said the sharing of idle computers is an aspect of what Condor can do, but it’s not the entire story. “It allows sharing of computing, and through the sharing you take advantage of idle computing, but you also allow the shifting of computing to where it’s needed most.”
Distributed or grid computing has for years enjoyed ample traction in academia, but it’s also gaining ground in corporate America. Campus researchers, particularly those in computer science departments, have been an excellent proving ground, which bodes well for the collaborative concept envisioned for southeastern Wisconsin, Livny said.
“At UWM, the LIGO [Scientific Collaboration Research Group] has been using Condor for a long time, so they have been collaborating with us,” he noted. “The idea is that though the Milwaukee Institute, the Milwaukee community can come together and establish shared distributed computing infrastructure for the local community by putting together the diverse resources that they have.”
Ultimately, the Milwaukee Institute intends to use the Condor software – a collaborative website is under development – but that’s only one of the collaborations being envisioned. The institute is in conversations with the National Center for Supercomputer Applications at the University of Illinois and the San Diego Super Computer Center at the University of California-San Diego, which have applications for scheduling and resource management services that are different than Condor.
In addition, the four Milwaukee institutions offer underused computer systems; in fact, some are only 20 percent utilized. One of the goals of the institute is to aggregate those systems to support existing projects and amortize their cost over a number of other users that now lack computational support.
“That argument has found an number of sympathetic ears, especially in CIO offices of the four institutions,” Bayne said. “They not only have embedded costs for the hardware, but for the power, systems programming, and systems management costs, and they’d like to see their ample investments better utilized.”
Business modeling
According to Byrnes, one of the private-sector benefits of the process will be to bring to the surface the scientific research that’s being done in the area so that businesses can determine whether they would work with a sponsor.
In addition, the institute is organizing six industry clusters – healthcare, manufacturing, financial services, transportation, defense, and media. The purpose of each cluster is to identify the research agenda common to all participants in that cluster, and then communicate those research needs to the local universities so they can begin to form related research projects – with the understanding that part of the funding would come from businesses interested in a given project.
Byrnes said the institute plans to use public sector research institution capacity for the foreseeable future. In the distant future, it may incorporate in the network some private-sector computers – which would raise complexity and security issues – but the idea is to network with the public sector community first.
It’s possible that the institute’s intellectual property will primarily be middleware that it hopes to develop as open-source software. However, some of the individual projects that it supports with computational resources may have intellectual property associated with them, and project participants would iron out those IP issues.
“What we hope to do is have our shared IP policy, where those who participate agree that the IP that’s developed is pre-competitive, and therefore they all have our shared, non-exclusive license to it,” Byrnes explained, noting that not all the details have been worked out yet. “We expect to handle the IP in a manner that’s consistent with the whole idea of the institute.”
While benefits could accrue to businesses, they will be developed in the lab. “This is an attempt to bring together the technical community in the area, rather than the executive community – people who love and understand technology as opposed to those who think of it as a cost center,” Byrnes said. “The long-term effect of it is to encourage innovation, and innovation doesn’t occur on a balance sheet and an income statement. It occurs in the laboratory.”
The new cyber infrastructure is being hosted at Tushaus Computing, a Milwaukee-based web hosting company.
Funding model
In funding the initial phases, Mason Wells is joined by Rockwell Automation, Johnson Controls, and Metavante, who have provided an undisclosed but significant contributions to get the project though its first two years. Long-term funding would come from foundations and government agencies like the National Science Foundation, the Department of Defense, and the National Institutes of Health.
Part of the motivation for doing this collaboratively is that more federal grant applications are based on the collaborative use of resources. “For southeastern Wisconsin to engage in activities that bring federally funded research, we need to go after research contracts,” said Bayne, “and they tend by nature to be multi-disciplinary and multi-institutional.”
Eventful week
Livny will deliver the keynote address when the institute holds its first public event, a day-long workshop on cyber infrastructure, on Thursday, March 13. The event will be the first of several “CI” days that will focus on high-performance computational systems that support collaborative research.
Livny emphasized that projects like Condor and the Milwaukee Institute are not so much about technology, but the collaboration the technology enables. “I think it’s important to communicate that the goal is not to push a technology,” he said, “but to push a collaborative concept.”
Related stories
• Rockwell’s $1M gift to UWM will support advanced automation research
• UW Condor project is a piece of CERN’s massive grid computer
• UW computer scientists tout achievements and explain industry shortcomings
• John Byrnes: We Need Techo-Savy Leaders – Don’t just blame government and education for economic problems
Relared Links
• John Byrnes: Let’s Get Serious
• John Byrnes: Lets Get Serious: Another Look
• John Byrnes: We Need Techo-Savy Leaders – Don’t just blame government and education for economic problems
• EthoStream will use Nanya Networks broadband products
• Two UW-Madison professors named to National LambdaRail Networking Research Council
• Madison grid computing efforts attack big research questions
• UW Condor project is a piece of CERN’s massive grid computer
• UW computer scientists tout achievements and explain industry shortcomings
• Landweber on networks, open source, and NSF budget cuts