I have a question about the vision of HCatalog and how it fits in with 'NoSQL' projects other than Hive and Pig.
Currently HCatalog enables sharing of data between Pig and Hive users on a common cluster. This fits the analytics workload when more casual users consume data in Hive that were created in Pig by a data scientist or programmer. There is another workflow that is common in analytics application development, where data is prepared in Pig or Hive, and published for consumption in a web application via a specialized store - key/value, document, etc. or even a relational database.