rowlike

ROWlike is a term used in data management to describe a row-oriented approach to data storage and processing. In this usage, data is organized so that complete records (rows) are stored contiguously on disk or in memory, and most operations access or manipulate entire rows rather than single columns. ROWlike systems emphasize transactional workloads where reads and updates often involve multiple fields of a single record.

Characteristics include efficient retrieval of full records, support for per-row updates and deletes, row-level compression or encoding, and indexing strategies that minimize record fetch time. Execution engines process data in row batches, enabling cache-friendly access that preserves the locality of all fields belonging to a row. This contrasts with columnar (column-wise) storage, which colocates values from the same column across many rows and is typically optimized for analytic scans.

ROWlike is often discussed in the context of hybrid storage architectures and NewSQL databases that aim to balance transactional performance with scalable analytics. Some experimental or vendor-provided engines offer a ROWlike mode that can be toggled to optimize for OLTP workloads, whereas other systems rely primarily on columnar or pure row-store designs.

Limitations: row-wise storage can suffer from lower compression ratios on wide schemas and increased I/O for analytical queries that touch a subset of columns. When used appropriately, ROWlike can provide fast, single-record transactions and straightforward data updates.

Related terms include row store, columnar store, OLTP, OLAP, and hybrid database architectures.