Re:Dropbox (YC S07) Announces 4 Million Users, Hires A VP Fr...
by glhaynes on Wednesday January 20, 2010 @05:50PM
(#1066021)
"Dropbox couldn't do that because each user would be searching a different data set (their own files)."
I don't really understand the claim here: they'd have fewer machines but also far fewer users and a much smaller (collection of) datasets.
Re:Dropbox (YC S07) Announces 4 Million Users, Hires A VP Fr...
by easp on Thursday January 21, 2010 @09:12PM
(#1068931)
Huh? So they have to store the inverted index for my files on disk, rather than in RAM and so it takes 1s to return a search result, rather than 0.2. With all the cold data they are storing, I imagine their servers have io ops to spare.
The fact that it's run on a server machine and not on a desktop machine has little effect on performance, and it probably doesn't make sense for Dropbox to provide the resources for good search when your desktop should be able to do a fine job of it.