Was reading a neat article over at The Register again about the new Google indexing system. This caught my eye:
“The TPC-E results suggest a promising direction for future investigation. We chose an architecture that scales linearly over many orders of magnitude on commodity machines, but we’ve seen that this costs a significant 30-fold overhead compared to traditional database architectures.
Kind of makes you think… I guess if your operating at the scale they are, the overhead is not a big deal, they’ll probably a find a way to reduce(ha ha, map reduce, get it? sorry) it over time.