AWARE: Adaptive Wide-Area Replication for Fast and Resilient Byzantine Consensus
Christian Berger, Hans P. Reiser, João Sousa, Alysson Bessani
备注：
This paper consists of 16 pages in total. This paper is the accepted version to be published in IEEE Transactions on Dependable and Secure Computing (2020). For the published version refer to DOI https://doi.org/10.1109/TDSC.2020.3030605
With upcoming blockchain infrastructures, world-spanning Byzantine consensus is getting practical and necessary. In geographically distributed systems, the pace at which consensus is achieved is limited by the heterogenous latencies of connections between replicas. If deployed on a wide-area network, consensus-based systems benefit from weighted replication, an approach that utilizes extra replicas and assigns higher voting power to well connected replicas. This enables more choice in quorum formation and replicas can leverage proportionally smaller quorums to advance, thus decreasing consensus latency. However, the system needs a solution to autonomously adjust to its environment if network conditions change or faults occur. We present Adaptive Wide-Area REplication (AWARE), a mechanism which improves the geographical scalability of consensus with nodes being widely spread across the world. Essentially, AWARE is an automated and dynamic voting weight tuning and leader positioning scheme, which supports the emergence of fast quorums in the system. It employs a reliable self-monitoring process and provides a prediction model seeking to minimize the system's consensus latency. In experiments using several AWS EC2 regions, AWARE dynamically optimizes consensus latency by self-reliantly finding a fast weight configuration yielding latency gains observed by clients located across the globe.
Byzantine state-machine replication (SMR) ensures the consistency of replicated state in the presence of malicious replicas and lies at the heart of the modern blockchain technology. Byzantine SMR protocols often guarantee safety under all circumstances and liveness only under synchrony. However, guaranteeing liveness even under this assumption is nontrivial. So far we have lacked systematic ways of incorporating liveness mechanisms into Byzantine SMR protocols, which often led to subtle bugs. To close this gap, we introduce a modular framework to facilitate the design of provably live and eff…
In this paper we demonstrate how stateful Byzantine Fault Tolerant services may be hosted on a Chord ring. The strategy presented is fourfold: firstly a replication scheme that dissociates the maintenance of replicated service state from ring recovery is developed. Secondly, clients of the ring based services are made replication aware. Thirdly, a consensus protocol is introduced that supports the serialization of updates. Finally Byzantine fault tolerant replication protocols are developed that ensure the integrity of service data hosted on the ring.
We present ezBFT, a novel leaderless, distributed consensus protocol capable of tolerating byzantine faults. ezBFT's main goal is to minimize the client-side latency in WAN deployments. It achieves this by (i) having no designated primary replica, and instead, enabling every replica to order the requests that it receives from clients; (ii) using only three communication steps to order requests in the common case; and (iii) involving clients actively in the consensus process. In addition, ezBFT minimizes the potentially negative effect of a byzantine replica on the overall system performance. W…
Distributed Learning often suffers from Byzantine failures, and there have been a number of works studying the problem of distributed stochastic optimization under Byzantine failures, where only a portion of workers, instead of all the workers in a distributed learning system, compute stochastic gradients at each iteration. These methods, albeit workable under Byzantine failures, have the shortcomings of either a sub-optimal convergence rate or high computation cost. To this end, we propose a new Byzantine-resilient stochastic gradient descent algorithm (BrSGD for short) which is provably robu…
We show that asynchronous $t$ faults Byzantine system is equivalent to asynchronous $t$-resilient system, where unbeknownst to all, the private inputs of at most $t$ processors were altered and installed by a malicious oracle. The immediate ramification is that dealing with asynchronous Byzantine systems does not call for new topological methods, as was recently employed by various researchers: Asynchronous Byzantine is a standard asynchronous system with an input caveat. It also shows that two recent independent investigations of vector $ε$-agreement in the Byzantine model, and then in the fa…
Traditional Byzantine resilient algorithms use 2f+1 vertex disjoint paths to ensure message delivery in the presence of up to f Byzantine nodes. The question of how these paths are identified is related to the fundamental problem of topology discovery. Distributed algorithms for topology discovery cope with a never ending task, dealing with frequent changes in the network topology and unpredictable transient faults. Therefore, algorithms for topology discovery should be self-stabilizing to ensure convergence of the topology information following any such unpredictable sequence of events. We pr…
We propose a novel robust aggregation rule for distributed synchronous Stochastic Gradient Descent~(SGD) under a general Byzantine failure model. The attackers can arbitrarily manipulate the data transferred between the servers and the workers in the parameter server~(PS) architecture. We prove the Byzantine resilience of the proposed aggregation rules. Empirical analysis shows that the proposed techniques outperform current approaches for realistic use cases and Byzantine attack scenarios.
Université Pierre et Marie Curie - Paris 6（巴黎第六大学皮埃尔与玛丽·居里大学）；LIP6-CNRS（巴黎第六大学计算机实验室（法国国家科学研究中心））
We propose the first deterministic algorithm that tolerates up to $f$ byzantine faults in $3f+1$-sized networks and performs in the asynchronous CORDA model. Our solution matches the previously established lower bound for the semi-synchronous ATOM model on the number of tolerated Byzantine robots. Our algorithm works under bounded scheduling assumptions for oblivious robots moving in a uni-dimensional space.
Byzantine fault tolerant (BFT) state machine replication (SMR) is an important building block for constructing permissioned blockchain systems. In contrast to Nakamoto Consensus where any block obtains higher assurance as buried deeper in the blockchain, in BFT SMR, any committed block is secure has a fixed resilience threshold. In this paper, we investigate strengthened fault tolerance (SFT) in BFT SMR under partial synchrony, which provides gradually increased resilience guarantees (like Nakamoto Consensus) during an optimistic period when the network is synchronous and the number of Byzanti…
This paper investigates the problem \textit{good-case latency} of Byzantine agreement, broadcast and state machine replication in the synchronous authenticated setting. The good-case latency measure captures the time it takes to reach agreement when all non-faulty parties have the same input (or in BB/SMR when the sender/leader is non-faulty). Previous result implies a lower bound showing that any Byzantine agreement or broadcast protocol tolerating more than $n/3$ faults must have a good-case latency of at least $Δ$, where $Δ$ is the assumed maximum message delay bound. Our first result is a …
Today's cyber-physical systems face various impediments to achieving their intended goals, namely, communication uncertainties and faults, relative to the increased integration of networked and wireless devices, hinder the synchronism needed to meet real-time deadlines. Moreover, being critical, these systems are also exposed to significant security threats. This threat combination increases the risk of physical damage. This paper addresses these problems by studying how to build the first real-time Byzantine reliable broadcast protocol (RTBRB) tolerating network uncertainties, faults, and att…
We present new protocols for Byzantine state machine replication and Byzantine agreement in the synchronous and authenticated setting. The celebrated PBFT state machine replication protocol tolerates $f$ Byzantine faults in an asynchronous setting using $3f+1$ replicas, and has since been studied or deployed by numerous works. In this work, we improve the Byzantine fault tolerance threshold to $n=2f+1$ by utilizing a relaxed synchrony assumption. We present a synchronous state machine replication protocol that commits a decision every 3 rounds in the common case. The key challenge is to ensure…
Byzantine fault tolerant protocols enable state replication in the presence of crashed, malfunctioning, or actively malicious processes. Designing such protocols without the assistance of verification tools, however, is remarkably error-prone. In an adversarial environment, performance and flexibility come at the cost of complexity, making the verification of existing protocols extremely difficult. We take a different approach and propose a formally verified consensus protocol designed for a specific use case: secure logging. Our protocol allows each node to propose entries in a parallel subro…
Docker container virtualization technology is being widely adopted in cloud computing environments because of its lightweight and effiency. However, it requires adequate control and management via an orchestrator. As a result, cloud providers are adopting the open-access Kubernetes platform as the standard orchestrator of containerized applications. To ensure applications' availability in Kubernetes, the latter uses Raft protocol's replication mechanism. Despite its simplicity, Raft assumes that machines fail only when shutdown. This failure event is rarely the only reason for a machine's malf…
The popularization of blockchains leads to a resurgence of interest in Byzantine Fault-Tolerant (BFT) state machine replication protocols. However, much of the work on this topic focuses on the underlying consensus protocols, with emphasis on their lack of scalability, leaving other subtle limitations unaddressed. These limitations are related to the effects of maintaining a durable blockchain instead of a write-ahead log and the requirement for reconfiguring the set of replicas in a decentralized way. We demonstrate these limitations using a digital coin blockchain application and BFT-SMaRt, …
Shandong University（山东大学）；University of Chinese Academy of Sciences（中国科学院大学）；University of California, Los Angeles（加利福尼亚大学洛杉矶分校）；City University of Hong Kong（香港城市大学）
Both Byzantine resilience and communication efficiency have attracted tremendous attention recently for their significance in edge federated learning. However, most existing algorithms may fail when dealing with real-world irregular data that behaves in a heavy-tailed manner. To address this issue, we study the stochastic convex and non-convex optimization problem for federated learning at edge and show how to handle heavy-tailed data while retaining the Byzantine resilience, communication efficiency and the optimal statistical error rates simultaneously. Specifically, we first present a Byzan…
The accelerated digitalisation of society along with technological evolution have extended the geographical span of cyber-physical systems. Two main threats have made the reliable and real-time control of these systems challenging: (i) uncertainty in the communication infrastructure induced by scale, and heterogeneity of the environment and devices; and (ii) targeted attacks maliciously worsening the impact of the above-mentioned communication uncertainties, disrupting the correctness of real-time applications. This paper addresses those challenges by showing how to build distributed protocols…
Stanford University（斯坦福大学）；University of Pennsylvania（宾夕法尼亚大学）；University of Wisconsin（威斯康星大学）
There has been a growing need to provide Byzantine-resilience in distributed model training. Existing robust distributed learning algorithms focus on developing sophisticated robust aggregators at the parameter servers, but pay less attention to balancing the communication cost and robustness. In this paper, we propose Solon, an algorithmic framework that exploits gradient redundancy to provide communication efficiency and Byzantine robustness simultaneously. Our theoretical analysis shows a fundamental trade-off among computational load, communication cost, and Byzantine robustness. We also d…
École Polytechnique Fédérale de Lausanne（洛桑联邦理工学院）
Byzantine resilience emerged as a prominent topic within the distributed machine learning community. Essentially, the goal is to enhance distributed optimization algorithms, such as distributed SGD, in a way that guarantees convergence despite the presence of some misbehaving (a.k.a., {\em Byzantine}) workers. Although a myriad of techniques addressing the problem have been proposed, the field arguably rests on fragile foundations. These techniques are hard to prove correct and rely on assumptions that are (a) quite unrealistic, i.e., often violated in practice, and (b) heterogeneous, i.e., ma…
International Institute of Information Technology Bangalore（班加罗尔国际信息技术学院）
This paper proposes a Robust Gradient Classification Framework (RGCF) for Byzantine fault tolerance in distributed stochastic gradient descent. The framework consists of a pattern recognition filter which we train to be able to classify individual gradients as Byzantine by using their direction alone. This filter is robust to an arbitrary number of Byzantine workers for convex as well as non-convex optimisation settings, which is a significant improvement on the prior work that is robust to Byzantine faults only when up to 50% of the workers are Byzantine. This solution does not require an est…
With the emergence of cross-organization attack-prone byzantine fault-tolerant (BFT) systems, so-called Blockchains, providing asynchronous state machine replication (SMR) solutions is no longer a theoretical concern. This paper introduces ACE: a general framework for the software design of fault-tolerant SMR systems. We first propose a new leader-based-view (LBV) abstraction that encapsulates the core properties provided by each view in a partially synchronous consensus algorithm, designed according to the leader-based view-by-view paradigm (e.g., PBFT and Paxos). Then, we compose several LBV…
Service replication distributes an application over many processes for tolerating faults, attacks, and misbehavior among a subset of the processes. The established state-machine replication paradigm inherently requires the application to be deterministic. This paper distinguishes three models for dealing with non-determinism in replicated services, where some processes are subject to faults and arbitrary behavior (so-called Byzantine faults): first, a modular approach that does not require any changes to the potentially non-deterministic application (and neither access to its internal data); s…
The problem of Byzantine consensus has been key to designing secure distributed systems. However, it is particularly difficult, mainly due to the presence of Byzantine processes that act arbitrarily and the unknown message delays in general networks. Although it is well known that both safety and liveness are at risk as soon as $n/3$ Byzantine processes fail, very few works attempted to characterize precisely the faults that produce safety violations from the faults that produce termination violations. In this paper, we present a new lower bound on the solvability of the consensus problem by d…
Existing approaches to tolerate Byzantine faults in geo-replicated environments require systems to execute complex agreement protocols over wide-area links and consequently are often associated with high response times. In this paper we address this problem with Spider, a resilient replication architecture for geo-distributed systems that leverages the availability characteristics of today's public-cloud infrastructures to minimize complexity and reduce latency. Spider models a system as a collection of loosely coupled replica groups whose members are hosted in different cloud-provided fault d…
We present two distributed algorithms for the {\em Byzantine counting problem}, which is concerned with estimating the size of a network in the presence of a large number of Byzantine nodes. In an $n$-node network ($n$ is unknown), our first algorithm, which is {\em deterministic}, finishes in $O(\log{n})$ rounds and is time-optimal. This algorithm can tolerate up to $O(n^{1 - γ})$ arbitrarily (adversarially) placed Byzantine nodes for any arbitrarily small (but fixed) positive constant $γ$. It outputs a (fixed) constant factor estimate of $\log{n}$ that would be known to all but $o(1)$ fracti…
We consider the message complexity of State Machine Replication protocols dealing with Byzantine failures in the partial synchrony model. A result of Dolev and Reischuk gives a quadratic lower bound for the message complexity, but it was unknown whether this lower bound is tight, with the most efficient known protocols giving worst-case message complexity $O(n^3)$. We describe a protocol which meets Dolev and Reischuk's quadratic lower bound, while also satisfying other desirable properties. To specify these properties, suppose that we have $n$ replicas, $f$ of which display Byzantine faults (…
Distributed machine learning algorithms enable learning of models from datasets that are distributed over a network without gathering the data at a centralized location. While efficient distributed algorithms have been developed under the assumption of faultless networks, failures that can render these algorithms nonfunctional occur frequently in the real world. This paper focuses on the problem of Byzantine failures, which are the hardest to safeguard against in distributed algorithms. While Byzantine fault tolerance has a rich history, existing work does not translate into efficient and prac…
Machine learning has begun to play a central role in many applications. A multitude of these applications typically also involve datasets that are distributed across multiple computing devices/machines due to either design constraints (e.g., multiagent systems) or computational/privacy reasons (e.g., learning on smartphone data). Such applications often require the learning tasks to be carried out in a decentralized fashion, in which there is no central server that is directly connected to all nodes. In real-world decentralized settings, nodes are prone to undetected failures due to malfunctio…
Distributed Computing Laboratory (DCL), École Polytechnique Fédérale de Lausanne (EPFL)（洛桑联邦理工学院（EPFL）分布式计算实验室（DCL））
Momentum is a variant of gradient descent that has been proposed for its benefits on convergence. In a distributed setting, momentum can be implemented either at the server or the worker side. When the aggregation rule used by the server is linear, commutativity with addition makes both deployments equivalent. Robustness and privacy are however among motivations to abandon linear aggregation rules. In this work, we demonstrate the benefits on robustness of using momentum at the worker side. We first prove that computing momentum at the workers reduces the variance-norm ratio of the gradient es…
A self-stabilizing is naturally resilient to transients faults (that is, faults of finite duration). Recently, a new class of protocol appears. These protocols are self-stabilizing and are moreover resilient to a limited number of permanent faults. In this article, we interest in self-stabilizing protocols that tolerate very hard permanent faults: Byzantine faults. We introduce two new scheme of Byzantine containment in self-stabilizing systems. We show that, for the problem of BFS spanning tree construction, the well known self-stabilizing protocol min+1 provides without significant modificat…
Microsoft Research Redmond（微软雷德蒙德研究院）；University of Waterloo（滑铁卢大学）；IST Austria（奥地利科学技术研究所）
We study adversary-resilient stochastic distributed optimization, in which $m$ machines can independently compute stochastic gradients, and cooperate to jointly optimize over their local objective functions. However, an $α$-fraction of the machines are $\textit{Byzantine}$, in that they may behave in arbitrary, adversarial ways. We consider a variant of this procedure in the challenging $\textit{non-convex}$ case. Our main result is a new algorithm SafeguardSGD which can provably escape saddle points and find approximate local minima of the non-convex objective. The algorithm is based on a new…
University of Southern California（南加州大学）；University of California, Riverside（加州大学河滨分校）
Secure federated learning is a privacy-preserving framework to improve machine learning models by training over large volumes of data collected by mobile users. This is achieved through an iterative process where, at each iteration, users update a global model using their local datasets. Each user then masks its local model via random keys, and the masked models are aggregated at a central server to compute the global model for the next iteration. As the local models are protected by random masks, the server cannot observe their true values. This presents a major challenge for the resilience o…
In this work, we consider the distributed stochastic optimization problem of minimizing a non-convex function $f(x) = \mathbb{E}_{ξ\sim \mathcal{D}} f(x; ξ)$ in an adversarial setting, where the individual functions $f(x; ξ)$ can also be potentially non-convex. We assume that at most $α$-fraction of a total of $K$ nodes can be Byzantines. We propose a robust stochastic variance-reduced gradient (SVRG) like algorithm for the problem, where the batch gradients are computed at the worker nodes (WNs) and the stochastic gradients are computed at the server node (SN). For the non-convex optimization…
Université Pierre et Marie Curie - Paris 6（巴黎第六大学（皮埃尔和玛丽·居里大学））；CNRS（法国国家科学研究中心）
Given a set of robots with arbitrary initial location and no agreement on a global coordinate system, convergence requires that all robots asymptotically approach the exact same, but unknown beforehand, location. Robots are oblivious-- they do not recall the past computations -- and are allowed to move in a one-dimensional space. Additionally, robots cannot communicate directly, instead they obtain system related information only via visual sensors. We draw a connection between the convergence problem in robot networks, and the distributed \emph{approximate agreement} problem (that requires co…
In this work, we consider the resilience of distributed algorithms based on stochastic gradient descent (SGD) in distributed learning with potentially Byzantine attackers, who could send arbitrary information to the parameter server to disrupt the training process. Toward this end, we propose a new Lipschitz-inspired coordinate-wise median approach (LICM-SGD) to mitigate Byzantine attacks. We show that our LICM-SGD algorithm can resist up to half of the workers being Byzantine attackers, while still converging almost surely to a stationary region in non-convex settings. Also, our LICM-SGD meth…
Generalization Error Analysis for Attack-Free and Byzantine-Resilient Decentralized Learning with Data Heterogeneity
School of Computer Science and Engineering, Sun Yat-Sen University（中山大学计算机科学与工程学院）；College of Computer, National University of Defense Technology（国防科技大学计算机学院）
Distributed consensus is a key enabler for many distributed systems including distributed databases and blockchains. Canopus is a scalable distributed consensus protocol that ensures that live nodes in a system agree on an ordered sequence of operations (called transactions). Unlike most prior consensus protocols, Canopus does not rely on a single leader. Instead, it uses a virtual tree overlay for message dissemination to limit network traffic across oversubscribed links. It leverages hardware redundancies, both within a rack and inside the network fabric, to reduce both protocol complexity a…
Traditional resilient systems operate on fully-replicated fault-tolerant clusters, which limits their scalability and performance. One way to make the step towards resilient high-performance systems that can deal with huge workloads, is by enabling independent fault-tolerant clusters to efficiently communicate and cooperate with each other, as this also enables the usage of high-performance techniques such as sharding and parallel processing. Recently, such inter-cluster communication was formalized as the Byzantine cluster-sending problem, and worst-case optimal protocols have been proposed t…
