promote the development of applications by users because of the flexibil- ity it gave developers.
applications to work, the IP address of each end point must be known by each end, something not possible with a NAT. NATs also break many other Internet protocols, such as end-2-end IPsec security, and even basic file transfers using FTP. While not prohibiting innovation, NATs do make it more difficult.
The end-2-end argument is moving into the mainstream of telecommu- nications policy because lawyers such as Larry Lessig [12][13] have real- ized the importance of end-2-end structure and its relationship to telecommunications policy. There are some important policy questions that the end-2-end argument can help resolve. Of particular interest to Lessig is policy related to owners of cable networks and independent ser- vice providers that want to provide network services using the basic infra- structure of the cable network. Should cable network owners be able to control what services (and hence what content) a user can choose from?
The end-2-end principle implies that this is not a good idea. Users must be able to choose the services and the providers of these services without interference from cable network owners. Any other policy stifles innova- tion. Today’s telecom policy about open cable network access will sculp- ture tomorrow’s landscape; we need to do it right. As the Internet shows,
“right” means allowing end users to innovate today with broadband Inter- net services and letting users, not cable network owners, choose the ser- vices and content they want.
Conclusion
The end-2-end argument has provided, and will continue to provide, insight into what network services the network infrastructure should offer.
The end-2-end theory is becoming more mature, allowing managers to understand when to apply the end-2-end argument in the context of how well they understand the needs of the users they are serving. End-2-end architecture has the most value when users have uncertain needs because it allows more experimentation. Knowing when there is value in letting users create their own services and when the cost of user innovation is too great allows management decisions that create the most value. The end-2- end principle is another tool for the network manager’s toolbox.
Understanding the end-2-end argument is one methodology that investors can apply to value investments in network-based services. It may be that the benefits of centralized management will far outweigh the value of the flexibility allowed by end-2-end architecture. Alternatively, it may be that uncertainty is so great, efficiency is of little value. Knowing when the 44 Chapter 3
value of distributed end-2-end architecture is worth more than cost-efficient centralized structure helps an investor decide when to invest in services based on their architecture and the level of market uncertainty.
The next chapter examines the difference between distributed (end-2- end) and centralized management. It provides more detail about the advantages of distributed management, such as easy innovation, and the disadvantages, such as inefficiencies in the use of resources. It further explains why centralized management is more efficient, and it offers other business advantages, such as protecting users from each other. It explains why centralized management is inflexible in the context of allowing users to innovate. It proposes a multilevel hierarchical structure to frame the process of deciding about how to provide network-based services within an organization. Examples are given from email, voice, and information services, illustrating how each of these services exists with many different management architectures, each structure meeting the needs of a particu- lar user group. Whereas these services are different, a common thread runs through the examples because of how market uncertainty affects the deci- sions about management structure. It illustrates the complexity of the deci- sion between centralized and distributed management structures.
Internet End-2-End Argument 45
47 Understanding the link between management architecture and market uncertainty is critical to successful decisions about how to build a network infrastructure and services using this infrastructure. Network-based ser- vices coexist with a continuum of management structure from centralized to distributed, as discussed in Chapter 2. It is expected that experimenta- tion will be easiest toward the distributed end of the spectrum and more difficult as the management structure becomes more centralized. Examples from email, voice services, and informational portal sites show that suc- cessful implementations of all these services coexist with different man- agement structures. The different management architectures appeal to users with different business needs. Companies with critical success fac- tors dependent on cutting-edge telecommunications services need more flexibility than centralized management allows, but organizations that have critical success factors less related to technology find centralized managed services a good fit with their needs because of its efficiency. The link between management structure and market uncertainty helps investors better structure their investment strategy and managers build network infrastructure and services creating the most value.