Technical Examination of Key-Value Store Scalability

Key value stores encapsulate data, metadata, a globally unique identifier and data attributes into a single immutable entity termed an object. In an object store, objects are organized in a flat address space i.e every object exists at the same level in a large scalable pool of storage. Key value stores follow location independent addressing and are accessed via the unique identifier. Key value stores are designed primarily to manipulate data sets that does not have a predefined data model or in other words is unstructured. The key operations for an object store include retrieval (get), store(put) and deletion(delete). In this survey, we aim to evaluate different classes of key values stores in terms of feature set, design choices, architectures, performance and their adaptability in the current market.