I was thinking we could use a job queue to process the facebook imports:
http://kr.github.com/beanstalkd/
or
http://www.danga.com/gearman/

Then we can use that to distribute processes for google sync, rss feed import, twits, backup import....