Both relational databases and NoSQL approaches can process data with billions of data points. It is difficult to make a meaningful and concrete recommendation with the information provided. It would be useful to know more about what you are trying to do with the data, what are your options regarding your equipment and network topology, etc.
I assume that since you are currently using a relational database, you have probably already reviewed the section or structured your large tables so that your query performance is satisfactory. This activity in itself may be nontrivial, but IMHO, a good database design with optimized sql, can go a very long way before there is a clear need to explore alternatives.
However, if your data usage looks like once, read it often, connection dependencies can be managed, and you need to perform some aggregations on the data set, then you can start looking for alternative approaches, such as Hadoop or MongoDB - however, this choice involves compromises from the point of view view of their performance, capabilities, platform requirements, latency, etc. Your specific question about integration between the NoSQL repository and the SQL database at the query level may not be implemented without some duplication of data between them. For example, MongoDB does not like joins (http://stackoverflow.com/questions/4067197/mongodb-and-joins), so you should develop your persistence model with this in mind, and this may be due to data duplication.
What I'm trying to do is determine the “right” approach will depend on your specific purpose and limitations.