Netty in Action

Foreword

There was a day when people believed that web application servers would let us forget about how we write HTTP or RPC servers. Unfortunately, this daydream did not last long. The amount of load and the pace of the functional changes we are dealing with continue to increase beyond the extent that a traditional three-tier architecture can afford, and we are being forced to split our application into many pieces and distribute them into a large cluster of machines.

Running such a large distributed system leads to two interesting problems—the cost of operation and latency. How many machines could we save if we improved the performance of a single node by 30%, or by more than 100%? How could we achieve minimal latency when a query from a ...