The Database Types Landscape

Relational (OLTP)

Examples: Postgres, MySQL, SQL Server.

VaultCommerce's system of record — customers, orders, payments, inventory. Fixed schema, JOIN queries, ACID transactions. The default choice when relationships and integrity dominate.

SELECT c.email, COUNT(o.id) AS order_count
FROM customers c
LEFT JOIN orders o ON o.customer_id = c.id
GROUP BY c.email;

Document stores

Examples: MongoDB, Couchbase.

Flexible JSON documents for entities with varying attributes. VaultCommerce stores supplier catalog feeds where each vendor sends different spec fields. Not ideal for multi-document transactions without extra design.

Key-value and in-memory

Examples: Redis, Memcached, DynamoDB (key-value access model).

O(1) lookups by key. VaultCommerce uses Redis for shopping-cart hashes and rate-limit counters. DynamoDB serves keyed access at AWS scale with partition-key design.

Wide-column / column-family

Examples: Cassandra, HBase, ScyllaDB.

Optimised for high-write, append-heavy workloads spread across nodes. VaultCommerce's analytics pipeline writes clickstream events partitioned by session_id and time bucket — millions of inserts per minute without updating existing rows.

Graph databases

Examples: Neo4j, Amazon Neptune.

Nodes and edges for relationship-heavy traversals. "Customers who bought X also bought Y" across deep recommendation chains — three-hop neighbour queries that explode in SQL JOINs are natural in Cypher:

MATCH (c:Customer)-[:PURCHASED]->(p:Product {sku: 'VLT-42'})
      -[:BOUGHT_WITH]->(rec:Product)
RETURN rec.name, count(*) AS strength
ORDER BY strength DESC LIMIT 10;

Search engines

Examples: Elasticsearch, OpenSearch.

Inverted indexes for full-text search, fuzzy matching, and aggregations. VaultCommerce's product search UI queries Elasticsearch; Postgres holds canonical product rows.

Time-series databases

Examples: TimescaleDB, InfluxDB, Prometheus.

Timestamp-indexed metrics and events. VaultCommerce tracks API latency and order-rate dashboards — append-only series with efficient range scans and downsampling.

Vector databases

Examples: pgvector (Postgres extension), Pinecone, Weaviate.

Store embedding vectors for similarity search. VaultCommerce's "shop by photo" feature compares uploaded image embeddings against a product vector index — nearest-neighbour queries impractical in plain B-tree indexes.

Data warehouses

Examples: Snowflake, BigQuery, Redshift.

Columnar analytics over historical data. Nightly ETL copies orders from Postgres into Snowflake for finance and merchandising reports — not for live checkout.

Choosing without catalogue shock