• Tidak ada hasil yang ditemukan

Information

Dalam dokumen Network Services Investment Guide (Halaman 82-89)

the first service provider to build a system that let its customers create and manage an online store. ViaWeb managed the servers and payment, while the store managed its content. This was a good match for what the market was looking for. No matter what degree of control the user has over con- tent, the outsourced solution means that the user does not need to manage the hardware and software of the server. The degree of market uncertainty helps determine the best structure for outsourcing and helps the user decide how much content, if any, he or she wants to manage.

Yahoo!’s main business is its Web site. It would never consider outsourc- ing its development or management. Because adding new services is Yahoo!’s lifeblood, control of its Web-based services is a critical success fac- tor and therefore must be under Yahoo!’s control. Organizations such as Yahoo! need more control and flexibility than any outsourced solution is capable of providing. Being able to decide when to upgrade to new ver- sions of HTML or HTTP is critical to Yahoo!’s success, and putting these decisions in the hands of others would be unwise for any organization with a competitive advantage that depends on the creativity and robust- ness of its Web site.

Figure 4.8 Outsourced information server.

Internet Access Company A

Company A's Content Who manages

Content?

Web Hosting site Web server Internet

Management Structure of Network-Based Services 67

Figures 4.9 and 4.10 illustrate the centralized and distributed structure possible with a Web-server. The centralized model depicted in Figure 4.9 consists of large servers, such as Sun’s Fire 15K, 12K, and Enterprise 10000 servers or IBM’s zSeries 900. Servers like this can cost over a million dol- lars. These systems are robust, and the vendors provide good support. This efficiency comes at a high price — it’s not easy to experiment with new ser- vices on large centralized servers. This difficulty of experimentation is due to the large number of users depending on each server. Managers of these large systems exercise extreme care when upgrading applications. The more distributed architecture in Figure 4.10 is far more flexible in terms of experimentation. In this distributed structure, Web requests are routed to one of many Web servers. No single Web server is critically important.

With this type of architecture, it is easy to change the software of one server. If the new software is bad, it is not a catastrophic event, as it would be with a large centralized server, because a smaller, more isolated group of users is affected. The centralized architecture in Figure 4.9 works best in low market uncertainty, and the distributed structure in Figure 4.10 meets the needs of users better when market uncertainty is high because it allows more experimentation.

Figure 4.9 Centralized in-house information server.

Internet Access Company A

Internet

Company A's Content Web server 68 Chapter 4

Figure 4.10 Distributed in-house information server.

How should organizations structure their internal Web server infra- structure? Reading the trade rags brings only confusion. Sun and IBM sug- gest that expensive large servers are the best architecture, but other vendors recommend using a greater number of less expensive servers based on the Intel architecture. Both groups present arguments that are logical. Large centralized servers are indeed easier and less expensive to manage and have extremely high reliability. Server farms of smaller sys- tems, while harder to manage, are easy to expand incrementally, have good reliability because of the redundancy of having many servers, and allow easy experimentation with new services. Yahoo!’s decision to build a dis- tributed infrastructure fits well with the high level of market uncertainty surrounding the Web-based applications it provides.

Yahoo!’s architecture of a server farm of Intel-based PCs running a type of Unix allows Yahoo! to experiment with new services very effectively. By installing a new service or changing an existing one on a single server, only a small fraction of users will see this new or changed service. If things don’t work, it’s easy to back out. On the other hand, if the experiment is a success,

Company A

Internet

Company A's Content Web server Company A's

Content Web server Company A's

Content Web server

Internet Access Web Load Sharing

Management Structure of Network-Based Services 69

it’s easy to phase in the new service. This allows Yahoo! to experiment with many new services and pick the best ones. It provides a way to gauge results because the new system can be compared to the old system in a con- trolled way. There are many trade-offs to consider when looking at server architecture, including flexibility, ease of experimentation, and business and technical advantages. In Yahoo!’s case, the degree of market uncer- tainty, coupled with the company’s needs, implies that the distributed architecture makes sense.

The choices managers make to provide Web-based informational ser- vices are illustrated in Figure 4.11, which shows the hierarchical structure of this decision. Similar to email and voice, the first choice to make is whether to outsource managing the Web server. Low market uncertainty implies that outsourcing this function might work well. Management must decide who manages the content of the Web site: the service provider or the user. On the other hand, if Web-server ownership and management is picked because of higher market uncertainty, then the next-level decision is about the architecture of the server infrastructure — should it be distrib- uted or centralized? As with email services and voice services, Web-based informational services fit nicely into this hierarchical framework where market uncertainty helps decide whether a distributed or centralized man- agement structure will create the most value.

Figure 4.11 Hierarchical structure of management architecture of informational services.

Outsource

Self-managed Web-server Info

SP manages

content

User manages

content

Central Server

Distributed Servers No

Yes

Centralized Outsource Web-server

Distributed Low

MU

High MU Centralized Distributed

Efficient Flexible

70 Chapter 4

There are good arguments on both sides of this issue about how to struc- ture a large-scale Web server installation, making the choice of architecture difficult. The answer again boils down to market uncertainty and the ability to meet user needs. High uncertainty implies that the value of experimenta- tion from the distributed model exceeds the benefits of consolidating many smaller servers into a centralized unit. When market uncertainty is low, then the ability to experiment is of less value because experiments have sim- ilar results. Low market uncertainty means you don’t need to experiment to see if particular users will accept or reject the service — you know the result of the experiment without performing it. Market uncertainty is a key factor in decisions pertaining to the choice of management structure.

Conclusion

This chapter discussed the management structure of network-based ser- vices in more detail. It explored why experimentation is so hard with a cen- tralized structure and why distributed management architecture promotes innovation through experimentation. It examined the hierarchical struc- ture of management architecture and how market uncertainty factors into the decision about management structure at each level. It discussed three different network-based services (email, voice, and informational) and illustrated different management structures for each of them and how they coexist. The discussion about email examined outsourcing compared to internal email management. It showed the flexibility of the distributed management solution compared to the easy management of the central- ized outsourced model. Next, voice services were discussed in the context of whether an organization should outsource its voice services with Cen- trex or buy and manage a PBX. It also looked at the advantages and disad- vantages of a centralized versus a distributed architecture of a PBX. Last, this chapter discussed how to provide information over the Web. Out- sourcing was compared to managing the Web server, and different Web server architectures were examined. The flexibility of the server farm model was compared to the efficiency of a large centralized server. These three examples illustrate how different management structures can coexist and how each structure meets the needs of some users best. It is the market uncertainty of the particular user that is important in the choice of man- agement architecture.

The next chapter examines the theory of options and its extension to real options, which form the basis of the ideas in this book. It takes the intuitive approach by leaving out the complex mathematics behind the theory and explains in practical terms how and why understanding options creates value from uncertainty. It is important because it is the foundation behind

Management Structure of Network-Based Services 71

the theory in this book about why high market uncertainty requires a man- agement structure that allows easy experimentation. Understanding how to think with an options point of view enables managers and investors to gain value from uncertainty — a powerful tool in today’s crazy world.

Options theory illustrates the value of managerial flexibility by showing the value of allowing managers to alter course as current conditions change. Understanding options thinking is a critical advantage in today’s uncertain world.

72 Chapter 4

73 Investors and managers that understand how to think in terms of options are at a competitive advantage in today’s uncertain world. Options were invented to help manage uncertainty in financial markets. The options way of thinking is critical to the modern manager as a strategy to profit from uncertainty. The greater the uncertainty, the greater the potential value of thinking with this options framework. The options point of view means a focus on flexibility and the ability to evaluate the current situation and alter the strategic path to better match the contemporary environment. The man- ager or investor thinking with the options mind set is not afraid of uncer- tainty because he or she understands how to craft opportunities that have increased value because of the uncertainty. Real options are the expansion of this theory to help manage non-financial assets in uncertain environ- ments. Learning how to create value from uncertainty is a good reason why understanding options is important to modern managers and investors.

This chapter examines the options theory and real options at an intuitive level. It illustrates how this theory helps managers make choices that max- imize value in uncertain conditions. It shows the link between uncertainty and value. Several applications of real options are presented: first the value of a staged investment strategy when building an Information Technology (IT) infrastructure and, second, the value of modularity in computer systems. This argument about the value of modularity is the underlying

Intuitive View of

Dalam dokumen Network Services Investment Guide (Halaman 82-89)