Byzantine disk paxos

Ittai Abraham, Gregory V. Chockler, Idit Keidar, Dahlia Malkhi
2004 Proceedings of the twenty-third annual ACM symposium on Principles of distributed computing - PODC '04  
We present Byzantine Disk Paxos, an asynchronous sharedmemory consensus protocol that uses a collection of n > 3t disks, t of which may fail by becoming non-responsive or arbitrarily corrupted. We give two constructions of this protocol; that is, we construct two different building blocks, each of which can be used, along with a leader oracle, to solve consensus. One building block is a shared wait-free safe register. The second building block is a regular register that satisfies a weaker
more » ... ation (liveness) condition than wait freedom: its write operations are wait-free, whereas its read operations are guaranteed to return only in executions with a finite number of writes. We call this termination condition finite writes (FW), and show that consensus is solvable with FW-terminating registers and a leader oracle. We construct each of these reliable registers from n > 3t base registers, t of which can be non-responsive or Byzantine. All the previous wait-free constructions in this model used at least 4t + 1 fault-prone registers, and we are not familiar with any prior FW-terminating constructions in this model. Algorithms, reliability Keywords Shared-memory emulations, Byzantine failures, termination conditions, consensus 1. One flavor is message-passing client-server systems in which servers store information on behalf of clients and the only communication is between clients and servers. Scalability is achieved by making servers as light as possible. Thus, the servers can be modeled as storage components. Examples of systems built using this approach include Fleet [24], SBQ-L [25], Agile Store [18], Coca [29], and [4]. 2. The second flavor is given by today's peer-to-peer systems. These systems consist of a collection of nodes spread all over the Internet that store data objects. Naturally, due to their Internet-wide deployment, the storage nodes are prone to malicious attacks. This motivates adopting a Byzantine failure model for the storage nodes. Examples of peer-to-peer systems that adopt storage-centric replication to support availability in face of Byzantine failures include Rosebud [26] and [27] . 3. The third flavor directly expresses an emerging network technology, the Storage Area Network (SAN).
doi:10.1145/1011767.1011801 dblp:conf/podc/AbrahamCKM04 fatcat:e6stixp2vram7nsacirgegv75m