Catalog
concept#Data#Platform#Architecture#Reliability

NoSQL Database

Non-relational database systems with flexible schemas, designed for horizontal scalability and varied consistency models.

NoSQL databases are non-relational storage systems that offer schema flexibility, horizontal scalability, and varied consistency models.
Established
Medium

Classification

  • Medium
  • Technical
  • Architectural
  • Intermediate

Technical context

ETL/streaming pipelines (e.g., Kafka, Dataflow)Search indexing systems (e.g., Elasticsearch)Backup and monitoring tools

Principles & goals

Design data model according to access patternsMake explicit decisions about consistency and replicationPlan operationalization (backup, monitoring, recovery) early
Build
Enterprise, Domain, Team

Use cases & scenarios

Compromises

  • Data inconsistencies with incorrect consistency configuration
  • Uncontrolled schema growth and data heterogeneity
  • Vendor lock-in due to proprietary queries or operational practices
  • Align data model to read and write routines
  • Document explicit replication and consistency policies
  • Integrate monitoring and alerting for latency, throughput, and errors

I/O & resources

  • Access patterns and expected load profiles
  • Data volume and growth forecasts
  • Requirements for consistency and latency
  • Decision for a NoSQL paradigm (document/column/key-value/graph)
  • Operationalized architecture with replication and backup
  • Metrics and alerts for operation and scaling

Description

NoSQL databases are non-relational storage systems that offer schema flexibility, horizontal scalability, and varied consistency models. They suit large, heterogeneous datasets, high write throughput, and distributed deployments. Typical decisions involve consistency versus availability, indexing strategies, backup approaches, and data modeling for application patterns.

✔

  • High horizontal scalability for large datasets
  • Schema flexibility enables rapid product iteration
  • Specialized models (key-value, document, column) for varied workloads

✖

  • Lack of universal ACID guarantees in many systems
  • Heterogeneous APIs and query features hinder portability
  • Operational complexity (sharding, replication, backups)

  • Throughput (write/read ops/s)

    Measures the number of successful operations per second and indicates scalability.

  • Latency (p95/p99)

    Indicates response time for critical read and write paths.

  • Error rate / success ratio

    Monitors stability and operationalization (e.g., write errors, replication failures).

Real-time analytics with event sourcing

A company stores user events in a document-based NoSQL database and runs derived aggregations for dashboards.

Product catalog as document collection

An online store uses a document-oriented system to efficiently represent variable product attributes and multilingual data.

Session store with key-value system

Web application uses a NoSQL key-value store as a fast central session storage for scalable frontends.

1

Analyze access patterns and data volume

2

Select a suitable NoSQL model and implement a prototype

3

Perform load and chaos testing

4

Operationalize: configure replication, backup, monitoring

5

Incremental migration and validation in production

⚠️ Technical debt & bottlenecks

  • Provisional data models without migration paths
  • Monolithic dependencies on specific NoSQL APIs
  • Insufficient tests for replication and failover scenarios
IndexingNetwork partitionsBackup/Recovery
  • Using a document DB for heavily relational transactions
  • Sharding without analyzing access patterns
  • Omitting monitoring and alerting in production
  • Underestimating operationalization costs
  • Ignoring hidden consistency issues during replication
  • Missing strategy for schema evolution
Understanding of distributed systems and CAP theoremExperience with data modeling in NoSQL systemsOperational knowledge of replication, sharding, and recovery
Scalability for large data volumesLatency requirements and throughputData consistency and availability
  • • Constraints imposed by chosen consistency models
  • • Budget for storage and network in cloud environments
  • • Compatibility with existing integrations