Traditional Public Key Infrastructures (PKI) have not lived up to their promise because there are too many ways to define PKIs, too many cryptographic primitives to build them with, and too many administrative domains with incompatible roots of trust. Alpaca is an authentication and authorization framework that embraces PKI diversity by enabling one PKI to “plug in ” another PKI’s credentials and cryptographic algorithms, allowing users of the latter to authenticate themselves to services using the former using their existing, unmodified certificates. Alpaca builds on Proof-Carrying Authorization (PCA) [8], expressing a credential as an explicit proof of a logical claim. Alpaca generalizes PCA to express not only delegation policies but also the cryptographic primitives, credential formats, and namespace structure needed to use foreign credentials directly. To achieve this goal, Alpaca introduces a method of creating and naming new principals which behave according to arbitrary rules, a modular approach to logical axioms, and a domain-specific language specialized for reasoning about authentication. We have implemented Alpaca as a Python module that assists applications in generating proofs (e.g., in a client requesting access to a resource), and in verifying those proofs via a compact 800-line TCB (e.g., in a server providing that resource). We present examples demonstrating Alpaca’s extensibility in scenarios involving inter-organization PKI interoperability and secure remote PKI upgrade.

A popular technique for reducing the bandwidth load on Web servers is to serve the content from proxies. Typically these hosts are trusted by the clients and server not to modify the data that they proxy. SSL splitting is a new technique for guaranteeing the integrity of data served from proxies without requiring changes to Web clients. Instead of relaying an insecure HTTP connection, an SSL splitting proxy simulates a normal Secure Sockets Layer (SSL) [7] connection with the client by merging authentication records from the server with data records from a cache. This technique reduces the bandwidth load on the server, while allowing an unmodified Web browser to verify that the data served from proxies is endorsed by the originating server.