2009/12/10 Anthony <[email protected]> > 2009/12/10 Iván Sánchez Ortega <[email protected]> > >> El Jueves, 10 de Diciembre de 2009, Jeffrey Johnson escribió: >> > This all jives with my experience doing this same sort of thing. >> > >> > Processing (reprojection and tiling) can also be offloaded to Amazon >> > as well by using many instances in parallel (the way I usually do it). >> > >> > Serving is not the issue, we just need disk and bandwidth, CPU and >> > Memory are not horribly important here. >> >> <crazyidea> >> >> Maybe we could do distributed computing for the reprojection, ti...@home >> style? Something like "aer...@home"? >> >> </crazyidea> >> > > One issue there is that there's a lot more data to download than > ti...@home. How long would it take to download the preprocessed data > compared to processing it? > > Also, how would we obtain the preprocessed data? Do we have to order it on > disk, or is there a way we might be able to have each node download it from > someone else's server? If the latter, ti...@home style might make sense, > but we'd need permission otherwise it could be seen as a DDOS attack. >
Along those lines, maybe something a little less distributed? Mail out 1-1.5 terabyte hard drives with an appropriate amount of raw data, have the sponsor process the data and mail it back?
_______________________________________________ talk mailing list [email protected] http://openaerialmap.org/mailman/listinfo/talk_openaerialmap.org
