
1 Deconstructing Blockchains: A Comprehensive Survey on Consensus, Membership and Structure Christopher Natoli, Jiangshan Yu∗ , Vincent Gramoli , and Paulo Esteves-Verissimo Abstract—It is no exaggeration to say that since the introduc- papers [53, 123, 124, 150, 155, 168, 193, 196, 204], wiki tion of Bitcoin, blockchains have become a disruptive technology documentations [116, 156, 180, 183] and websites [5, 11, that has shaken the world. However, the rising popularity of the 130, 133]. This is in contrast with the traditional academic paradigm has led to a flurry of proposals addressing variations and/or trying to solve problems stemming from the initial research that conveys results through scientific publications specification. This added considerable complexity to the current typically peer-reviewed by specialists of particular domains, blockchain ecosystems, amplified by the absence of detail in many like distributed systems, cryptography, networking or game accompanying blockchain whitepapers. theory [63, 84, 94, 120]. The lack of detail has left room for Through this paper, we set out to explain blockchains in a interpretation that has sometimes led researchers to different simple way, taming that complexity through the deconstruction of the blockchain into three simple, critical components common conclusions [22, 75, 102]. to all known systems: membership selection, consensus mechanism Through this paper, we set out to explain blockchains in a and structure. We propose an evaluation framework with insight simple way, taming that complexity through the deconstruc- into system models, desired properties and analysis criteria, using tion of the blockchain into three simple, critical components the decoupled components as criteria. We use this framework to common to most known systems: membership selection, con- provide clear and intuitive overviews of the design principles behind the analyzed systems and the properties achieved. We sensus mechanism, and structure. The membership selection hope our effort will help clarifying the current state of blockchain component determines a committee of nodes that participate proposals and provide directions to the analysis of future pro- in the consensus; the consensus mechanism component is posals. responsible for deciding on the next block, run by the selected committee of nodes; and the structure component represents how the data is organized in the blockchain. Thanks to our I.I NTRODUCTION deconstruction, we are able to provide a clear and unique N 2008 Satoshi Nakamoto released Bitcoin [150], high- landscape of blockchains, as depicted in Figure 1. We then I lighting the power and importance of distributed systems. propose an evaluation framework with insights into different The use of distributed systems in our lives has been trans- key system models, desired properties, and analysis criteria, parently controlled by significant actors. The introduction using the decoupled components as parameters. We use this of Bitcoin provided the public with insight into ungoverned framework to provide overviews of the design principles distributed systems and sparked a movement towards decen- behind the analyzed systems and the achieved properties. tralization. Initially, the blockchain technology went vastly There are notable works aggregating and analyzing unnoticed and was heavily integrated with the deep web due to blockchains, each providing unique evaluation but primarily its pseudonymity properties [55, 217]. However, the concept focusing on one aspect and constructing specific frameworks to of the blockchain quickly gained interest and grew to what comparatively analyze. Cachin and Vukolic´ present a compar- arXiv:1908.08316v1 [cs.DC] 22 Aug 2019 it is today. The growing interest promoted the introduction ison of consensus protocols in permissioned blockchains [49]. of new chains and the growth of distributed ledger tech- Abraham and Malkhi also present the idea of deconstructing nologies [4, 10, 14, 78, 137, 141, 170]. However, the rising the blockchain into layers [18] but with a focus on relating popularity of the paradigm has led to a flurry of proposals Nakamoto’s consensus versus Byzantine fault tolerant (BFT) addressing variations and/or trying to solve problems arising protocols. Similarly, Vukolic´ [212] explores the contrast be- from the initial specification. tween blockchains and BFT replicated state machines, pointing The complexity introduced through popularity was am- to scalability problems faced by both and provide insight into plified by the absence of detail in the proposals. As an upcoming proposals. Gramoli [97] investigates mainstream example, the vast majority were presented through white blockchains and discusses classical Byzantine consensus in the context of the blockchain and discusses some of the dangers ∗ Corresponding author. of misunderstanding guarantees, problems that we address and Christopher Natoli is with University of Sydney, Australia. Vincent Gramoli is with University of Sydney and CSIRO Data61, systematize in our paper. Wang et al. [213] approach the Australia. blockchain consensus mechanisms in the perspective of game Jiangshan Yu is with Monash University, Australia. E-mail: theory and the strategies of adoption of nodes. Bano et al. [26] [email protected] Paulo Esteves-Verissimo is with the Interdisciplinary Centre for Security, provide an overview on how different blockchain consensus Reliability and Trust, University of Luxembourg, Luxembourg. work, and compare existing Proof-of-* against other proposed 2 Figure 1. The blockchain landscape. consensus mechanisms based on their properties. However, no vides background information about blockchains and the core other work has deconstructed the blockchain into these unique properties relating to our decomposition. Section III outlines components and analysed all three. the criteria used for comparison and analysis. Section IV dis- While existing work provides good analysis of specific sets cusses and analyzes the membership selection and proposals. of the properties of blockchains, it remained challenging, even Section V provides an analysis of consensus mechanisms. for the educated but non-expert readers to get a thorough Section VI overviews structure proposals for the blockchain. and comparative picture of the design principles of different Section VII highlights proposed attacks relating to the mem- complicated systems. To our knowledge, we are the first to bership selection and consensus. Section VIII discusses other classify blockchain based on a decomposition into membership insights and aspects. We conclude with a summary of our selection, consensus mechanism and structure, categorizing analysis and discuss future directions. and providing an analysis of leading proposals. As an addition to our analysis, we provide an overview of the attacks and II.B ACKGROUND threats related to membership selection, consensus goals, and In this section, we explore the foundational concepts of the structural assumptions. We hope that our analysis will clarify blockchain and discuss the initial specification of Bitcoin. We future design directions and inspire new designs based on in- then discuss the applicability of blockchains and conclude by novative coherent combinations revealed by our decomposition highlighting the known impediments. and categorization. Thus, this paper presents the following contributions: A. Blockchain • It innovatively deconstructs the blockchain into three Blockchain is an append-only distributed ledger of trans- simple, critical components: membership selection, con- actions. First introduced with Bitcoin [150], the blockchain sensus mechanism and structure. originated as a decentralized electronic payment system, which • It provides an evaluation framework with insight into removed the need for any third-party involvement for payment system models, desired properties and analysis criteria, transfers. The original Bitcoin blockchain organized data as using the decoupled components as parameters. a chain of blocks directed by utilizing the block hash val- • It provides a clear and intuitive overview of the design ues, hence the name “blockchain”. Later proposals expanded principles behind analyzed systems and their properties, from a single chain to parallel chains [120, 209, 221] and in terms of the parameters above. graphs [25, 128, 168]. This has led to the term “blockchain” • It proposes, based on the analyzed state of the art, becoming a misnamed concept for distributed ledgers. a categorization of membership selection and consen- The decentralized nature of blockchains means that all sus approaches. In addition, it provides generic charts nodes verify and store the transactions that have taken place representing the design principles for each category of in the system, and propose new blocks to append to the membership selection, to simplify future designs. chain. The blockchain structure can be seen as a linearly The remainder of the paper is as follows. Section II pro- increasing linked list of transactions batched into blocks. 3 The chain begins with a genesis block at index 0 and each of the claimed output. An output has two fields, namely block appended links to its direct predecessor forming the value and address. They define the value to be transferred chain. This, however, is the combination of pre-existing ideas to the address. Bitcoins are just transaction outputs with an constructed together to form what the blockchain is today. arbitrary value with 8 decimal
Details
-
File Typepdf
-
Upload Time-
-
Content LanguagesEnglish
-
Upload UserAnonymous/Not logged-in
-
File Pages43 Page
-
File Size-