Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.538686
Title: Building a better grid authentication system with Kerberos
Author: McBride, David William
ISNI:       0000 0004 2705 0753
Awarding Body: Imperial College London
Current Institution: Imperial College London
Date of Award: 2011
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
Authentication is the act of verifying that the agent you are communicating with really is the same one you believe you're talking to. This capability is a vital prerequisite for security in many systems, electronic or otherwise. Given the existence of a trusted third party, we can effectively authenticate users and systems in a single locale using Kerberos. Across much larger networks, such as the Internet, we can instead use public keys certified by one of many trusted third parties. Current Grid systems use GSI, a system based on public keys, for the authentication of users and services alike. However, end-users are generally not proficient at handling such keys; they require a support mechanism, or security suffers. We describe how existing technologies - namely, Kerberos 5, PKCROSS and DNSSEC- bis - can be integrated to support a more usable and featureful authentication infrastructure suitable for a range of public Internet operations, including Grid applications, with fewer central points of failure (and attack), that retains the desirable security and performance attributes of existing approaches, while minimising the costs of deployment, upkeep and maintenance. We begin by describing specific use-cases of interest and present a set of requirements that a large-scale authentication mechanism should exhibit. We show that the Kerberos 5 cross-realm security model, given some additional Realm authentication mechanism, would provide an efficient and reliable authentication infrastructure that satisfies our requirements. We then show how to revise the proposed PKCROSS extensions for Kerberos to exploit the burgeoning DNSSEC-6is infrastructure to establish point-to-point cross-realm trust relationships autonomously, thereby enabling foreign services to authenticate a user via their home institutional Kerberos credentials. However, DNSSEC-6zs is unacceptably flawed: it introduces a single, ultimately trusted root authority for public Internet security operations. Therefore, we present a modified scheme which distributes this root authority between the operators of the top-level domains, whom we believe would collectively be more trustworthy - that retains the performance and much of the conceptual simplicity of the current design.
Supervisor: Darlington, John ; Knottenbelt, William ; Thornley, David Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.538686  DOI:
Share: