登入選單

返回Google圖書搜尋

Understanding Internet Naming: From the Modern DNS Ecosystem to New Directions in Naming

Tom Callahan

出版	Case Western Reserve University, 2013
URL	http://books.google.com.hk/books?id=tO-KAQAACAAJ&hl=&source=gbs_api

註釋

In this dissertation we study current and proposed implementations of Internet naming schemes. Creating reliable naming systems requires an up-to-date understanding of the current system's real-world operation. Therefore, our first goal is to understand the modern client-side Domain Name System (DNS) ecosystem through empirical study of both system components and operational DNS traffic. Next, we discuss two naming scenarios where the DNS is insufficient and propose solutions. One problem we examine--the second component of this dissertation--is that Internet transactions need a well-known rendezvous point to establish communication, often a DNS name. However, the static nature of these rendezvous points introduces brittleness to the process. Controversial rendezvous points (such as IP addresses and hostnames used to bootstrap into peer-to-peer networks) are often targeted by censors. Many factors can prevent the usage of a central hub in the absence of any adversary--central hubs can also be vulnerable to network failures, power failures, and human error. Rendezvous schemes based upon DNS are additionally vulnerable to failures of authoritative DNS servers and lapses in domain registration. Therefore, we design and evaluate a system that allows users to communicate without any centralized hub or fixed rendezvous point. Another problem we examine--and the third thrust of this dissertation--is that DNS does not encourage user-to-user information sharing in general. Publishing DNS records remains limited to systems administrators through often manual processes. Furthermore, we note that in general the DNS is used to name hosts, which most users are not interested in. Rather, users typically share content and pointers to other people on different Internet services; however, no DNS types exist to publish the Uniform Resource Locator (URL) for a user's webpage or an Instant-Messaging screen name used by a user. Therefore, we discuss a new naming system centered around users, allowing for secure publication and consumption of names by users and their applications.