Having recovered from the holidays and several proposals and reports, it is time for met to get back to blogging. Certainly lots of interesting things to write about.
Where better to start than a report of the world's biggest grid computer. Or maybe not ... I read a report from the Davos Forum that quotes Vint Cerf as saying that between 100 and 150 million of the world's 600 million computers are (unwittingly) engaged in botnets. The report notes that "a single botnet at one point used up about 15% of Yahoo's search capacity [retrieving] random text snippets to camouflage messages so that its spam e-mail could get past spam filters." Scary.
From a security perspective, I think it's a little bit of a stretch to call a botnet a grid since the resources owners (the legitimate ones that is) really do not get a fair chance to have input with regards to how their resources are used in a botnet. But a botnet is an interesting example of setting up a grid with crude but effective mechanisms for establishing policy and common purpose across a range of distributed resources. Since the botnet operators have little investment in the resources, they have less concern with security than most grid operators.
Posted by: Von | January 26, 2007 at 09:29 AM
What can we learn from running software on somewhat heterogeneous systems of widely varying hardware capabilities? Can software and OS, as in this case, or OS extensions, abstract the hardware disparateness sufficiently to provide a sufficiently homogeneous view to the compute job? Is the approach to throw enough compute nodes on a problem, and work in the greater design with the possibility of failure of some a more practical approach? Also note that they don't have queues to hassle them.
Posted by: Jens | January 27, 2007 at 11:14 AM