Skip to content
Book Open access

POSTER: Auditable and Fault-Tolerant Choreographic Cloud Services via Distributed Effects

Nov 2026 · 0 citations

Abstract

Data clean rooms (DCRs) are the cloud industry's answer to confidential multi-party data sharing under regulations such as GDPR and the EU Data Act, yet their trustworthiness hinges on hand-written distributed code: developers must weave audit logging and fault recovery into message-passing protocols without breaking the confidentiality and deadlock-freedom properties that make the protocols trustworthy in the first place. Choreographic programming derives the distributed implementation from one global protocol description, guaranteeing deadlock-freedom by construction, but existing frameworks hide local side-effects behind an opaque effect layer, leaving critical concerns such as auditing and fault recovery outside any formal guarantee. We present DEChor, a choreographic framework that lifts local effects into first-class distributed effects through a generic expose/embed interface, so a single global program can orchestrate communication, accumulate per-participant audit trails, and recover from per-participant failures without subverting the choreographic discipline. We prove that endpoint projection remains deadlock-free in the presence of distributed effects, and we deploy a healthcare-style DCR choreography on the Google Kubernetes Engine with up to 100~participants---to our knowledge the first deployment of a choreographic program on a public cloud---measuring that the distributed-effect machinery adds no measurable runtime overhead. The poster presents the design, formal guarantees, and the cloud deployment evaluation of DEChor, and invites discussion on threat models and auditing architectures for regulated multi-party computation services.

Read PDF

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.