What are some of the challenging problems to adapt to map reduce framework?

I have developed around 20 jobs on map reduce including the pagerank algorithm. I never found any challenging pro开发者_Go百科blems to adapt to mapreduce framework online. I would like to improve my skills. Is there such a guide?

---

What you are looking for is Data intensive programming task. Here is a similar question already posted at StackOverflow. I thought of suggesting this project because corpus from wikipedia is easily available, but as you can see it is already in progress.

Run a squid reverse proxy and collect the logs of those over a period of time. Now use those logs and try to derive meaningful interpretation of those and store them in suitable database for querying. This could be good project to do.