On Global Virtual Data Structures
Gian Pietro Picco, Amy Lynn Murphy, Gruia-Catalin Roman · 2020
In distributed computing, global information is rarely available and most actions are carried out locally. However, when proving system properties we frequently turn to defining abstractions of the global state, and when programming we often find it convenient to think of a distributed system as a global centralized resource. In this paper we build upon this observation and propose the notion of global virtual data structures as a model for building a new generation coordination models and middleware that allows programmers to think of local actions as having a global impact and places upon the underlying system the burden of preserving this appearance. The model itself is inherently peer-to-peer, lending itself toward applications which are largely decentralized, and built out of autonomous components.