Scalable transactions in the cloud: partitioning revisited

Abstract

Cloud computing is becoming one of the most used paradigms to deploy highly available and scalable systems. These systems usually demand the management of huge amounts of data, which cannot be solved with traditional nor replicated database systems as we know them. Recent solutions store data in special key-value structures, in an approach that commonly lacks the consistency provided by transactional guarantees, as it is traded for high scalability and availability. In order to ensure consistent access to the information, the use of transactions is required. However, it is well-known that traditional replication protocols do not scale well for a cloud environment. Here we take a look at current proposals to deploy transactional systems in the cloud and we propose a new system aiming at being a step forward in achieving this goal. We proceed to focus on data partitioning and describe the key role it plays in achieving high scalability.