Building The Grid
By Steve Wallage, Tue Oct 05 08:15:00 GMT 2004
Grid computing is a big IT buzzword, but will it make any difference to the mobile world?
Grid computing is hot. The term has been applied so loosely that you might think all IT was becoming “gridified”. In reality, the vast majority of grid applications are still taking place in the academic environment and replacing supercomputer processing.
Put simply, grid computing is a standardized version of distributed computing with non-trivial qualities of service. Perhaps the most famous example is SETI@Home that uses the spare cycles of PC users in its search for the next ET.
One segment which has shown a lot of interest in grid computing has been the telcos, particularly in Europe and Japan. Interestingly, the US telcos are lagging behind and show more interest in trying to apply the word grid to their own IP infrastructure. European carriers see grid as interesting to both their internal processes such as billing, and in the potential for external offerings. BT, Deutsche Telekom and Telecom Italia are leading the way here.
Although a number of the major telcos have mobile operations, it is clear that is the fixed side that is leading grid deployments. Vodafone has joined some of the grid consortia, including the European group CGrid, and is working with Oracle 10g to offer Oracle business services on the Vodafone network. However, it does not have a dedicated grid team. In terms of internal usage, MVNO Virgin Mobile has been analyzing the potential of grid in its billing arm.
But the interest from the mobile community looks set to grow as research develops into the potential for wireless grids. Three projects are leading the way.
Intel Research at Berkeley has been looking at how to link wireless devices to create a wireless grid. Intel's long-term view is that grids will be the mechanism for connecting all digital devices. They will enable utility computing, then autonomic computing (an approach to self-managed computing systems), and will be based on commodity architectures.
The Intel view is that distributed robotics will be enabled by small, low-cost flexible wireless devices with a flexible, open operating system and environment to combine sensing, communication and computations.
Metcalfe’s law states that grid-based resources become more valuable as the number of devices and users increases. The wireless grid makes it easier to extend grid computing to large numbers of devices that would otherwise be unable to participate and share resources. This can include all types of mobile devices, with a particular appeal for sensors, which can supply information on temperature, health, or pollution levels. Such sensors do not have the same constraints on processing power and battery limitations of wireless handsets.
TeleCom City, an economic and technology development project located north of Boston that’s funded by the National Science Foundation and led by Syracuse University and supported by other universities including MIT, Tufts and Boston University, with vendor involvement from BT, Cisco and four start-ups, is undertaking two-year project looking at how to use idle computing power in a wireless grid environment.
The project is examining some of the key challenges to wireless grids including security, standards and implementation challenges. On the standards side, there is a strong focus from researchers (as well as the fixed telcos) to use Web Services as the basis. On the security side, TeleCom City is exploring the use of the Friend-of-a-Friend (FOAF) protocol. In terms of implementation, a key challenge is that grids are typically based on continuously connected devices while mobile devices will need to dynamically enter and leave the grid. One assumption from most of the research projects is that the wireless grids will need to be connected to a fixed grid. TeleCom City researchers are also looking at how to overcome some of the battery and processing power limitations of mobile handsets.
A third project is at Syracuse University, where the Wireless Grids research team is building a project known as DARC* (Dark Star). The system lets wireless devices with no prior knowledge of each other collectively record and mix an audio signal such as a concert, speech, lecture or emergency event. The project demonstrates the potential of wireless grids and distributed ad hoc resource sharing to harness the combined ability of mobile devices in social contexts outside the expected environments for computing.
Many of these projects remain very immature, but this is true of grid applications in all commercial environments. It does seem that wireless sensors make far more sense as a possible wireless grid project than handsets, given the latter's limitations. It does not take a great leap of imagination to see the possible benefits of linked wireless sensors to collect and process information, and there are no doubt other research projects that are looking at this exact possibility.
Equally, a lot of the mobile operators are starting to ask their R&D departments to look at potential possibilities of grid. In fact, at least one wireless operator is looking at a “grid lite” offering that could connect handsets to a wired grid. The business model could allow users a reduced monthly rental if they donate cycles to the operator.
There are many obstacles to this, and it is at least several years away, but expect to start hearing a lot more about the potential for wireless grids.