Browse Books

Reviewer: Jeffrey B. Putnam

In any complex design and development process, there will be individuals who specialize in different areas. It can help this process if these specialists know something about the other specialties involved. If an expert in one field does not understand all of the problems involved, he or she may easily make design decisions that would be impossible to implement. In the worst case, this can result in the failure of a product; in less severe cases, it may involve serious delays, as the design gets reworked to accommodate the needs of all of the specialists. This can easily be avoided: put only those to work on a project who have expertise in all of the fields required. This, though, is generally impractical and probably overkill. It will usually suffice to ensure that all involved have at least a taste of knowledge in the various areas and that they are willing to learn more as needed (not, in itself, a trivial undertaking if the areas are fairly disparate). This book tries to do something like that for online game designers and network engineers?giving game designers some information about how networks work and network engineers some information about how games use the network. The book starts with an overview (chapters 2 and 3) of historical and recent online games and game systems, distinguishing between gaming consoles, handheld consoles, and personal computer games. Chapter 4 continues with a discussion of common network protocols, including Internet protocol, transmission control protocol (TCP), user datagram protocol (UDP), network address translation (NAT), dynamic host configuration protocol (DHCP), and domain name system (DNS). Network latency and jitter and loss patterns are then examined in chapters 5 and 6, along with a discussion of the implications these have for online games, techniques for overcoming latency (prediction, time warping), and some "visual tricks" for hiding network latency from the players. Making a game playable despite adverse network traffic patterns is then discussed (chapter 7), as is the interaction of network conditions and cheats (and cheat mitigation techniques). Chapter 8 returns to network architecture to discuss the properties of broadband access. How players come to a game server is covered in chapter 9, and the patterns of network use that arise are covered in chapter 10. Chapter 11 discusses future directions, including the impact that quality of service-oriented protocols might have on the online gaming business. Information on setting up a game server makes up chapter 12. Chapter 13 contains a collection of other resources. This all sounds rather promising, with the possible exception of some organizational issues. For instance, why is the coverage of broadband in chapter 8 not placed with the other network discussions earlier on__?__ And indeed, there is good information here. There are lots of good graphs for interesting statistics. For instance, one nice graph plots "frag" rate versus median latency. However, it is on the same page as a cumulative plot of played games versus median latency, and the fact that it is a cumulative plot seems odd?why not use the raw frequency instead__?__ As you read more, and especially in the chapter on game traffic patterns, the questions mount. In one plot of Quake III Arena packet size, packets with odd length seem to dominate packets with even length (packets of length 70 seem to be about two or three percent of the total, packets with length 71 are seven to nine percent of the total, and packets with length 72 again drop to about three percent). No explanation for this is given and the phenomenon is not even noted in the text. Perhaps some smoothing would have hidden this oddity and helped to illuminate other patterns. Many of the plots show interesting patterns, but in many of them the noise might have been reduced by some smoothing so the patterns could be made more evident. Ping and traceroute are mentioned as a way to gain some insight into network behavior, but increasingly UDP pings (at least) are being filtered by firewalls and so may be of less use than is claimed. TCP-based traceroute programs are not even mentioned. The "Lan Party" as a way for game players to get together on a (usually constructed for the occasion) fast local network is not mentioned; this is particularly odd, as it could have been a very productive source of statistics on client and server behavior. There are also a few places where some interesting (and perhaps questionable) statements are made. For example, on page 102 we read that "a game-hosting company, an Internet Service Provider, and a game manufacturer are aiming at the same things¿satisfied consumers," but the recent debate about net neutrality seems to indicate that these interests and those of the consumers may not be entirely congruent. There are also a few claims that games deserve special consideration by network service providers¿and little or no mention is made of potential competitive services (voice over Internet protocol, video conferencing, on-demand video downloading, and others that may emerge). It has been said that inside every thin person, there is a fat person struggling to get out. Similarly, inside this thin and undoubtedly interesting, but more often frustrating, book, there is a revealing and fascinating book waiting to be let out. I'll be looking for it when it is. Online Computing Reviews Service