Abstract
Multi-master replication in a distributed system setting allows each node holding a replica to update and query the local replica, and disseminate updates to other nodes. Obtaining high availability typically entails allowing replicas to diverge and requires a background mechanism for re-establishing consistency. Conflict-free Replicated Data Types (CRDTs) extend standard sequential data-Types with appropriate merge functions, and often can be composed together to create more complex ones. In this work we add a generic CRDT composition approach that explores the single-writer principle. By carefully controlling which part of the composition can be updated by each replica, we can derive efficient designs that cover new usecases. After introducing the new construction we exemplify some uses, including how to emulate a simple Doodle functionality for selecting a common meeting schedule among different participants.

Abstract
The reactive paradigm recently became very popular in user-interface development: updates - such as the ones from the mouse, keyboard, or from the network - can trigger a chain of computations organised in a dependency graph, letting the underlying engine control the scheduling of these computations. In the context of the Internet of Things (IoT), typical applications deploy components in distributed nodes and link their interfaces, employing a publish-subscribe architecture. The paradigm for Distributed Reactive Programming marries these two concepts, treating each distributed component as a reactive computation. However, existing approaches either require expensive synchronisation mechanisms or they do not support pipelining, i.e., allowingmultiple "waves" of updates to be executed in parallel. We propose Quarp (Quality-Aware Reactive Programming), a scalable and light-weight mechanism aimed at the IoT to orchestrate components triggered by updates of data-producing components or of aggregating components. This mechanism appends meta-information tomessages between components capturing the context of the data, used to dynamically monitor and guarantee useful properties of the dynamic applications. These include the so-called glitch freedom, time synchronisation, and geographical proximity. We formalise Quarp using a simple operational semantics, provide concrete examples of useful instances of contexts, and situate our approach in the realm of distributed reactive programming.

Abstract
This paper discusses the motivation and the challenges for providing a systematic and transparent approach for dealing with cross-system consistency. Our high level goal is to provide a way to avoid violations of causality when multiple systems interact, while (a) avoiding the redesign of existing systems, (b) minimizing the overhead, and (c) requiring as little developer input as possible.

Abstract
Programming models for building large-scale distributed applications assist the developer in reasoning about consistency and distribution. However, many of the programming models for weak consistency, which promise the largest scalability gains, have little in the way of evaluation to demonstrate the promised scalability. We present an experience report on the implementation and largescale evaluation of one of these models, Lasp, originally presented at PPDP '15, which provides a declarative, functional programming style for distributed applications. We demonstrate the scalability of Lasp's prototype runtime implementation up to 1024 nodes in the Amazon cloud computing environment. It achieves high scalability by uniquely combining hybrid gossip with a programming model based on convergent computation. We report on the engineering challenges of this implementation and its evaluation, specifically related to operating research prototypes in a production cloud environment.

Abstract
Flow-Updating (FU) is a fault-tolerant technique that has proved to be efficient in practice for the distributed computation of aggregate functions in communication networks where individual processors do not have access to global information. Previous distributed aggregation protocols, based on repeated sharing of input values (or mass) among processors, sometimes called Mass-Distribution (MD) protocols, are not resilient to communication failures (or message loss) because such failures yield a loss of mass. In this paper, we present a protocol which we call Mass-Distribution with Flow-Updating (MDFU). We obtain MDFU by applying FU techniques to classic MD. We analyze the convergence time of MDFU showing that stochastic message loss produces low overhead. This is the first convergence proof of an FU-based algorithm. We evaluate MDFU experimentally, comparing it with previous MD and FU protocols, and verifying the behavior predicted by the analysis. Finally, given that MDFU incurs a fixed deviation proportional to the message-loss rate, we adjust the accuracy of MDFU heuristically in a new protocol called MDFU with Linear Prediction (MDFU-LP). The evaluation shows that both MDFU and MDFU-LP behave very well in practice, even under high rates of message loss and even changing the input values dynamically.

Abstract
Keeping replicated data strongly consistent is convenient when communication is fast and available. In internet-scale distributed systems the reality of high communication latencies and likelihood of partitions, leads developers to adopt more relaxed consistency models, such as eventual consistency. Conflict-free Replicated Data Types, bring structure to the design of eventually consistent data management solutions, by precisely describing the behaviour under concurrent updates and guarantying a path to reconciliation. This paper offers a survey of the mathematical structures that support state based multi-master replication with reconciliation, and shows how state structures and state transformations can be composed to provide data types that are now used in practice in many geo-replicated systems.