An Aggregation Framework, Data Center Awareness functionality and improvements to concurrency are the major features in the latest release of MongoDB.
The aggregation framework is designed to make it easier to manipulate and process documents inside of MongoDB without needing to use Map-Reduce or separate application processes for data manipulation. Map-Reduce isn’t ideally suited for simple aggregation tasks such as totaling or averaging field values. The developers of MongoDB say the aggregation framework provides similar functionality to GROUP BY and related SQL operators as well as simple forms of self joins. The aggregation framework can also be used to reshape the returned data, adding computed fields, creating new virtual sub-objects, and extracting sub-fields into the top level of results.
Another improvement to the new version is better management of larger data sets that are more geographically distributed. The “Data Center Awareness” features include standard handling of read preferences across all drivers and sharded interfaces. Another feature designed for distributed clusters is tag aware sharding. This means you can ensure that data in a geographically distributed sharded cluster is always closest to the application that will use that data the most.
The other main changes to version 2.2 are in the area of concurrency. The global lock has been eliminated from the mongod process, and locking is now enforced per database. Locking has also been strengthened with the addition of a new subsystem that avoids locks under most page-fault events, and when writing to secondaries applications handle parallelism better.
There is a video showing the concurrency changes.