Exploring 10+ Database Types: From Relational to Vector and Beyond

Special‑Purpose Databases

Search Engine Databases

Designed for full‑text search, these systems store data using an inverted index that maps each term to the list of document identifiers containing that term. Query processing includes tokenization, stop‑word removal, and relevance scoring (e.g., TF‑IDF or BM25). Popular engines are Elasticsearch, Apache Solr, and the underlying library Apache Lucene.

Document Databases

Document stores keep semi‑structured JSON‑like documents. Schemas are flexible: new fields can be added without altering existing collections. Queries use a JSON‑based query language and can index nested fields. Leading products are MongoDB and Couchbase.

Time‑Series Databases

Optimized for timestamped data, they index on the time column and provide efficient range queries, down‑sampling, and aggregation functions (e.g., avg, max, percentile). Storage engines differ: InfluxDB uses a Time‑Structured Merge Tree (TSM) with B+‑tree time indexes; TimescaleDB extends PostgreSQL with hypertables that partition data by time intervals. Visualization is commonly done with Grafana.

Vector Databases

Store high‑dimensional vectors (e.g., embeddings) and support similarity search using distance metrics such as Euclidean, cosine, or inner product. Index structures include inverted file (IVF), hierarchical navigable small world graphs (HNSW), and KD‑trees. Notable implementations are Milvus, Pinecone, Faiss, and PostgreSQL’s vector extension.

Spatial Databases

Handle geometric objects (points, lines, polygons) and geographic coordinate systems. Spatial indexes (R‑tree, Quadtree) accelerate range and nearest‑neighbor queries. PostGIS adds spatial types and functions to PostgreSQL.

Graph Databases

Model data as nodes and edges, enabling efficient traversal queries (e.g., shortest path, neighborhood). Internally they use adjacency lists or matrices and may build graph‑specific indexes. Popular systems include Neo4j and TigerGraph.

Column‑Store Databases

Store data column‑wise rather than row‑wise, which improves compression and speeds up analytical (OLAP) queries that touch only a subset of columns. Examples are ClickHouse, Apache HBase, and Druid.

Multi‑Model Databases

Support multiple data models (relational, document, graph, key‑value) within a single engine, reducing architectural complexity for applications that need diverse storage. ArangoDB and OrientDB are representative multi‑model platforms.