Fast Distributed Transactions and Strongly Consistent Replication for OLTP Database Systems

Abstract

As more data management software is designed for deployment in public and private clouds, or on a cluster of commodity servers, new distributed storage systems increasingly achieve high data access throughput via partitioning and replication. In order to achieve high scalability, however, today's systems generally reduce transactional support, disallowing single transactions from spanning multiple partitions.This article describes Calvin, a practical transaction scheduling and data replication layer that uses a deterministic ordering guarantee to significantly reduce the normally prohibitive contention costs associated with distributed transactions. This allows near-linear scalability on a cluster of commodity machines, without eliminating traditional transactional guarantees, introducing a single point of failure, or requiring application developers to reason about data partitioning. By replicating transaction inputs instead of transactional actions, Calvin is able to support multiple consistency levels-including Paxos-based strong consistency across geographically distant replicas-at no cost to transactional throughput.Furthermore, Calvin introduces a set of tools that will allow application developers to gain the full performance benefit of Calvin's server-side transaction scheduling mechanisms without introducing the additional code complexity and inconvenience normally associated with using DBMS stored procedures in place of ad hoc client-side transactions.

Rethinking serializable multiversion concurrency control

“…In order to plan execution correctly, the concurrency control phase needs advance knowledge of each transaction's write-set. This requirement is not unique to BOHM -several prior systems exploit a priori information about transactions's read-and/or writesets [3,12,34,25]. These previous systems have shown that even though they need transactions' write-(and sometimes also read-) sets in advance, it is not necessary for transactions to pre-declare these read-/write-sets.…”

“…These previous systems have shown that even though they need transactions' write-(and sometimes also read-) sets in advance, it is not necessary for transactions to pre-declare these read-/write-sets. For example, Calvin proposes a speculative technique which predicts each transaction's read-/write-sets on the fly [34]. Furthermore, Ren et.…”

“…Traditional cursor-oriented database access, where transactions are submitted to the database in pieces, are therefore not supported. Furthermore, the write-set of a transaction must be deducible before the transaction begins -either through explicit write-set declaration by the program that submits the transaction, or through analysis of the transaction by the database system, or through optimistic techniques that submit a transaction for a trial run to get an initial guess for its write-set, and abort the transaction if the trial run resulted in an incorrect prediction [34,30].…”

Esext3 for Reducing the Effect of Harboring Bug in File System

“…ClvinFS is the scalable and robust enough to handle the distributed database system standard. This file system has the ability to work with semantics that can including fully linear by the help of random writes by handling concurrent users for the flow of random byte offsets as theses are in same file and it may be located across wide geographic areas [10] [11]. EXPLODE can comprehensively test storage systems by acclimatizing key designs from model scrutiny in a way that keeps their authority but removes their intrusiveness.…”

Tolerância a falhas para sistemas legados: um estudo de caso no exército brasileiro

“…Esta pesquisa foca nas características relacionadas ao armazenamento de arquivos, enquanto que os aspectos necessários para a replicação da aplicação e do banco de dados não serão aprofundados. Muito pode ser feito para tornar transparente a hospedagem distribuída da aplicação, por exemplo como apresentado nos trabalhos de Metz [39] e de Sarat et al [42], enquanto que a replicação de banco de dados pode ser alcançada com uso das técnicas apresentadas por Thomson et al [47] e por Chairunnanda et al [9].…”