Yes.

Yes, but only if one of the tables fits into memory

Yes, so long as both tables fit into memory.

No, MapReduce cannot perform relational operations.

No, but it can be done with either Pig or Hive.

Intermediate data in streamed across the network from Mapper to the Reduce and is never written to disk.

Into in-memory buffers on the TaskTracker node running the Mapper that spill over and are written into HDFS.

Into in-memory buffers that spill over to the local file system of the TaskTracker node running the Mapper.

Into in-memory buffers that spill over to the local file system (outside HDFS) of the TaskTracker node running the Reducer

Into in-memory buffers on the TaskTracker node running the Reducer that spill over and are written into HDFS.

Ingest the server web logs into HDFS using Flume.

Write a MapReduce job, with the web servers for mappers, and the Hadoop cluster nodes for reduces.

Import all users' clicks from your OLTP databases into Hadoop, using Sqoop.

Channel these clickstreams inot Hadoop using Hadoop Streaming.

Sample the weblogs from the web servers, copying them into Hadoop using curl.

Single point of failure in the NameNode.

Resource pressure on the JobTracker.

HDFS latency.

Ability to run frameworks other than MapReduce, such as MPI.

Reduce complexity of the MapReduce APIs.

Standardize on a single MapReduce API.

hadoop "mapred.job.name=Example" MyDriver input output

hadoop MyDriver mapred.job.name=Example input output

hadoop MyDrive ""D mapred.job.name=Example input output

hadoop setproperty mapred.job.name=Example MyDriver input output

hadoop setproperty ("mapred.job.name=Example") MyDriver input output