There are certainly situations where other database solutions will perform better. For example, PostgreSQL is missing features needed to perform well on some of the more difficult queries in the TPC-H test suite (see Chapter 8, Database Benchmarking, for more details). It's correspondingly less suitable for running large data warehouse applications than many of the commercial databases. If you need queries along the lines of some of the very heavy ones TPC-H includes, you may find that databases such as Oracle, DB2, and SQL Server still have a performance advantage worth paying for. There are also several PostgreSQL-derived databases that include features making them more appropriate for data warehouses and similar larger systems. Examples include Greenplum, Aster Data, and Netezza.
For some types of web applications, you can only get acceptable performance by cutting corners on the data integrity features in ways that PostgreSQL just won't allow. These applications might be better served by a less strict database, such as MySQL or even a really minimal one, such as SQLite. Unlike the fairly mature data warehouse market, the design of this type of application is still moving around quite a bit. Work on approaches using the key/value-based NoSQL approach, including CouchDB, MongoDB, and Cassandra, are all becoming more popular at the time of writing this. All of them can easily outperform a traditional database, provided you have no need to run the sort of advanced queries that key/value stores are slower at handling. PostgreSQL also natively supports and indexes the Json data type for a NoSQL data approach.
