I read the blog
tried to understand the architecture of the HBase. There is one thing that
makes me confusing. If I understand right, the client must connect the
Zookeeper cluster to get the metadata of particular table information and
then use this information to retrieve the table.
My question is the node which hosts the ".META. table" and "-ROOT- region"
is the node which belongs to HBase cluster or Zookeeper cluster? If they
belongs to the Zookeeper cluster, what is the functionality of the Master
node in the HBase cluster, only assigning the work node when starting the
HBase? In my point of view, the Master node will maintain the ".META. table"
and "-ROOT- region" information.