7.7 Concluding Thoughts

From a historical point of view, network performance has lagged behind system performance. The network has traditionally been a bottleneck to performance; you could always build processors fast enough to fill network pipes, and disk arrays faster than networks. One of the most interesting developments of the last five years has been the vast increase in available network throughput: from 1991 to 2001, local area network bandwidth across Ethernet has increased from 10 megabits a second to an even 1 gigabit a second -- and work on a future 10 Gb standard is ongoing. As available capacity keeps scaling up, systems are having an increasingly difficult time processing data fast enough to keep up with the data flow.

In this chapter, we've gone over a lot of material that systems administrators often aren't exposed to: the principles behind network operations, the physical media used to convey information over distances, the underlying interfaces (Ethernet, FDDI, ATM) and protocols (particularly the ever-tunable TCP), finishing up with a discussion of the applications that let us share files across the office or across the world. The theories here are complex; this chapter may require several readings . If this work interests you, pick up a few of the texts we've mentioned in this chapter. They can only improve your understanding of the interface between the very different fields of systems and network engineering.