Network systems, designed to support packet processing applications efficiently, must meet two, often-conflicting, requirements:

1. Support a large number of high-bandwidth links, and hence large system throughputs
2. Offer a wide range of flexible, upgradeable services (such as VPN, intrusion detection, and differentiated services) in addition to conventional bridging and forwarding functions

To meet simultaneously the demands of high performance and flexibility, network processors (NPUs) with designs optimized for processing packets at high rates have emerged. NPUs, and more generally the systems designed using NPUs, contain multiple heterogeneous processors, complex memory subsystems, and a non-uniform inter-processor communication architecture.

Unfortunately, the methodologies and tools needed for programming such network systems effectively are in their infancy. Today, to program such systems, programmers are required to be intimately familiar with the hardware; programmers develop hand-tuned code that manages carefully a variety of network system resources to ensure that the system can process incoming traffic at link speeds. This approach is tedious, slow, and yields code that is difficult to port from one network system to another. With the next-generation of network systems supporting a larger number and types of processor cores, more diverse memory hierarchies, and more complex processor interconnects, the difficulty in programming network systems will only increase over time.

The success in developing and deploying multi-service networks using programmable components depends critically on the design of a programming environment that simplifies software development for network systems. The objective of Project Shangri-La is to design a unified, easy-to-use programming environment that facilitates rapid development of portable, high-performance packet processing applications on NPU-based network systems.