In this talk we describe our work, at IBM Research on Project Kittyhawk, exploring the viability of a Global Scale Public Computer. We begin with a brief discussion of what we mean by a global scale public computer and global computation. We define our notion of a Global Scale Computer as a well specified public, "software-less", massively parallel system, on which users can construct services, of arbitrary scale within resource limits, out of metered and billed common units of its capacity grouped in domains of communication they specify and control. The basic approach taken in project Kittyhawk is the combining of global computation with a massively parallel processor.
The core of the talk presents our prototype system built on IBM's Blue Gene/P hardware platform. We describe the hardware and how we utilize it to permit principals to construct both private and shared computational environments from hardware-based common units of capacity in the form of Blue Gene/P nodes. These nodes are composed of cores, memory, and communication resources. To construct the environments we prototype support for control channels and a primitive we call a communication domain that establishes which nodes can communicate with each other. We describe some examples that utilize open-source software to construct internet-inspired global computation scenarios out of the raw hardware nodes and network topologies realized on communication domains.
After presenting the prototype we briefly discuss why, in contrast to trends in cloud computing, we have chosen to focus on hardware capacity rather than virtualization, and conclude by raising the question of how we want our digital future to evolve given the warnings and advice of pioneers such as John McCarthy and others.
Slides:
Download slides for this presentation in PDF format.
About the speaker:
Jonathan is a Research Scientist at IBM's T.J. Watson Research Center in New York. Jonathan received his Ph.D in Computer Science from the University of Toronto in 2005. His work focused on scalable systems software for large-scale, general purpose multi-processors. Specifically, he worked on an object model for systems software construction which permits and encourages fine grain control of sharing and attendant communication costs. He first pursued this work in the University of Toronto, Tornado operating system and then in the IBM K42 operating system. Jonathan's current research interest is in exploring structure in the complete execution of a modern computer including the stochastic interactions with the software, data and exogenous events. He is interested in how the structure can be defined, quantified and exploited. In 2007, along with his colleagues Volkmar Uhlig and Amos Waterland, he established Project Kittyhawk to explore conjectures about Global Scale computers and computation.
Contact information:
Jonathan Appavoo
T.J. Watson Research Center
1101 Kitchwan Road
Yorktown Heights, NY 10579
jappavoo@us.ibm.com