Aerospike Architecture Overview Whitepaper
This paper describes how the Aerospike architecture is derived from its core principles – NoSQL scalability and flexibility, along with traditional database consistency, reliability and extensibility. These principles are demonstrated in the shared-nothing distribution architecture, the schema-less data framework, the insistence on immediate consistency and atomicity, and system-wide fault-tolerance. In addition, the architecture is geared to operational efficiency, both in its ease of use for application developers and system administrators, and in its speed and low resource overhead requirements when running on off-the-shelf Linux environment. Aerospike represents a best of all worlds solution for applications requiring fast transactional access to high scale data.
ACID Support in Aerospike
This whitepaper defines provides a brief technical overview of ACID support in Aerospike. It includes a definition of ACID (Atomicity, Consistency, Isolation, Durability), and an overview of the CAP Theorem, which postulates that only two of the three properties of consistency, availability, and partition tolerance can be guaranteed in a distributed system at a specific time. Although Aerospike is an AP system with a proven track record of strong consistency with no downtime, this paper describes Aerospike’s unique approach to avoiding network partitioning to also ensure high consistency. In addition, the paper describes how Aerospike will give users the option to support a CP configuration with complete consistency in the presence of networking partitioning by reducing availability.
Proceedings of VLDB
Aerospike founders first presented this innovative architecture at the High Performance Transactions System conference and the VLDB (Very Large Databases) conference in 2010.