A MapReduce Dropbox

During the recent OSDC meeting there was a lot of discussion about a (hopefully) "viral" application or approach to aid scientists by allowing them to use OSDC resources. dropbox was a frequently occurring example. Taking the direct approach, I prototyped a simple map-reduce box, which behaves like a dropbox but it can also accept special files or folders representing map-reduce jobs. All file/folder referencing is done in local terms, thus "cloudifying" the storage and map-reduce service. As map-reduce is a fairly straightforward programming paradigm and hadoop resources can be secured at low cost, my intuition is that a map-reduce box could be immediately useful to scientists on its own or as part of a larger toolset. In this talk I will briefly introduce and demo the map-reduce dropbox. For the remainder of the time I hope we have a couple of brainstorming sessions, one about the potential of a map-reduce dropbox as a research component, within the current research activities of the group, and one about the map-reduce box as a potential product within or outside OSDC.