Migration in the Stencil Pluralist Cloud Architecture Tai Liu Zain Tariq Tencent America LLC New York University Abu Dhabi
[email protected] [email protected] Barath Raghavan Jay Chen University of Southern California International Computer Science Institute
[email protected] [email protected] ABSTRACT There are many important technical design challenges in decen- A debate in the research community has buzzed in the background tralized infrastructure, including security, naming, and more. Our for years: should large-scale Internet services be centralized or de- goal is to narrow the focus to the key issues the architecture must centralized? Now-common centralized cloud and web services have adjudicate as opposed to an individual application or service. We downsides—user lock-in and loss of privacy and data control—that argue that we need a pluralist architecture: one that allows the are increasingly apparent. However, their decentralized counter- co-existence of applications and seamless migration between them. parts have struggled to gain adoption, suffer from their own prob- Not only can such an architecture prevent user lock in, but it can lems of scalability and trust, and eventually may result in the exact also ease the pain of developing decentralized applications. Put same lock-in they intended to prevent. another way, a pluralist architecture is one that picks no winners: In this paper, we explore the design of a pluralist cloud architec- instead, it allows a marketplace of services to be developed, and ture, Stencil, one that can serve as a narrow waist for user-facing provides enough scaffolding and restrictions to ensure that the services such as social media.