05 — Library · 12 stacks

Stacks

Reference technology stacks by layer, with the technologies in each and the patterns they follow. Not endorsements.

Search stacksSort

Stacks

Showing 9 of 12 stacks

TS-100
Convention-driven monolith
Web application
One full-stack framework with strong conventions, a relational database, a cache, a background job runner and server-rendered pages enhanced with small amounts of JavaScript. One deployable, one team, most of the product built in a year.
Framework
Ruby on Rails8.1 · EOL 2027-10-10
Database
PostgreSQL18 · EOL 2030-11-14
Cache and queues
Redis8.10
Background jobs
Sidekiq
Front-end interactivity
Hotwire
Fits · Small teams building a business application end to end; the reliable choice when nobody can justify a distributed system yet.
patterns DP-240 DP-070 DP-060
- BC-720.20.10Application Development
- BC-720.10.10Solution Architecture
TS-050
Cross-platform mobile
Mobile
One TypeScript codebase that ships native iOS and Android applications, with a managed build and over-the-air update service, so a web-fluent team can deliver mobile without two native teams.
Framework
React Native0.87
Tooling and builds
Expo
Language
TypeScript
Navigation
React Navigation
Release automation
Fastlane
End-to-end testing
Detox
Fits · Product teams with React experience whose app is mostly forms, lists and content; graphics-heavy or hardware-bound apps still want native code.
patterns DP-120 DP-390 DP-020
- BC-720.20.10Application Development
- BC-1220.40.20App Store Submission
TS-060
Enterprise integration layer
API & integration
A gateway in front of published APIs, a routing and transformation engine for the system-to-system flows, a broker for asynchronous work, and a contract standard tying them together. The layer that lets a packaged-application estate change one system without renegotiating every interface.
API gateway
Kong Gateway3.9
Routing and mediation
Apache Camel4.22 · EOL 2027-08-11
Message broker
RabbitMQ4.3
Event streaming
Apache Kafka4.3
API contracts
OpenAPI Specification
Event contracts
AsyncAPI
Fits · Organizations with many packaged and legacy systems that must exchange data; the integration team becomes a product team or the layer becomes the new bottleneck.
patterns DP-110 DP-160 DP-170 DP-200 DP-140 DP-040
- BC-720.40Integration & API Delivery
- BC-1240.10API Design & Governance
TS-020
Event-driven microservices on Kubernetes
Event streaming
Independently deployable services on a container orchestrator, talking through a durable event log rather than direct calls, with a mesh handling service-to-service security and traffic policy. The shape most large product organizations converge on once a monolith stops scaling with the team.
Orchestration
Kubernetes1.37 · EOL 2027-10-28
Event log
Apache Kafka4.3
Service mesh
Istio1.31 · EOL 2027-02-28
Packaging
Helm
Continuous delivery
Argo CD3.5
Service framework
Spring Boot4.1 · EOL 2027-07-31
Schema governance
Confluent Schema Registry
Fits · Organizations with many teams and real traffic; the platform cost only pays back when team autonomy and independent release are the bottleneck.
patterns DP-130 DP-290 DP-180 DP-280 DP-430 DP-310 DP-010
TS-040
Machine learning platform
Machine learning
Experiment tracking, a model registry, a feature store and a serving layer on top of the container platform, so a model moves from a notebook to production through the same path every time and the lineage from training data to deployed version is recorded.
Language
Python3.15 · EOL 2031-10-31
Training framework
PyTorch
Experiment tracking and registry
MLflow
Pipelines
Kubeflow
Feature store
Feast
Serving
KServe
Orchestration
Kubernetes1.37 · EOL 2027-10-28
Fits · Teams running more than a handful of models in production; a single model is better served by a managed endpoint than by this much platform.
patterns DP-410 DP-460 DP-450
TS-030
Open lakehouse analytics platform
Data platform
Raw, refined and curated data as open-format tables on object storage, transformed by a distributed engine and a SQL modeling tool, scheduled by a workflow orchestrator. Analytics and machine learning read the same tables, which removes the warehouse-versus-lake split.
Object storage
Amazon S3
Table format
Apache Iceberg
Processing engine
Apache Spark4.2 · EOL 2028-01-11
SQL modeling
dbt2.0 · EOL 2027-09-14
Orchestration
Apache Airflow3.3
Query engine
Trino
Change capture
Debezium
Fits · Data teams that want engine choice and open formats over a single vendor warehouse, and have the engineering depth to run the pieces.
patterns DP-330 DP-300 DP-320
- BC-750.30Data Engineering & Pipelines
- BC-750.20.20Data Platform Design
TS-070
Open observability stack
Observability
Vendor-neutral instrumentation feeding separate stores for metrics, logs and traces, queried from one dashboard tool and alerted on by one rule engine. Owning the stack trades a license bill for operations work and data volume to manage.
Instrumentation
OpenTelemetry
Metrics
Prometheus3.15 · EOL 2026-11-06
Logs
Grafana Loki3.7
Traces
Grafana Tempo
Dashboards
Grafana13.2 · EOL 2027-05-18
Alert routing
Alertmanager
Fits · Platform teams that can run stateful services and want to avoid per-host or per-GB pricing; small teams are usually better off with a hosted service.
patterns DP-450 DP-460 DP-470
- BC-1230.20Observability
- BC-730.60.30Observability Platform
TS-120
Platform infrastructure baseline
Infrastructure
Cloud accounts and networks declared as code, a managed container platform per environment, a pull-based delivery controller reconciling clusters from Git, and secrets and certificates issued automatically. The foundation an internal platform team offers before any product team deploys.
Infrastructure as code
Terraform1.16
Container platform
Kubernetes1.37 · EOL 2027-10-28
GitOps delivery
Argo CD3.5
Secrets
HashiCorp Vault2.1
Certificates
cert-manager1.21
Policy
Open Policy Agent
Fits · Organizations running several product teams on shared cloud infrastructure; one team on one cluster can start with the managed provider's defaults.
patterns DP-410 DP-400 DP-350 DP-510
TS-010
Server-rendered React on serverless
Web application
A React framework that renders on the server by default, deployed to a serverless edge platform, with a managed Postgres behind it. The default for a small team shipping a content-heavy product fast.
Framework
Next.js16
Language
TypeScript
Styling
Tailwind CSS4.3
Database
PostgreSQL18 · EOL 2030-11-14
Hosting
Vercel
Fits · Small teams, content and catalog products, anything where time-to-first-release matters more than control of the runtime.
patterns DP-020
- BC-720.20Software Engineering
- BC-1210Platform Engineering
TS-090
Serverless API on AWS
API & integration
HTTP endpoints as individual functions behind a managed gateway, with a key-value store and a queue for asynchronous work, all declared as infrastructure code. No servers to patch and a bill that tracks requests, at the cost of cold starts and a vendor-shaped design.
Compute
AWS Lambdajava17.al2023 · EOL 2029-08-31
Gateway
Amazon API Gateway
Database
Amazon DynamoDB
Queue
Amazon SQS
Infrastructure as code
Terraform1.16
Language
TypeScript
Fits · Spiky or low-volume APIs, event handlers and integrations; steady high-throughput services usually cost less on containers.
patterns DP-110 DP-060 DP-180 DP-410 DP-440
- BC-1210.10.40Serverless & Edge Runtime
- BC-720.40.10API Design & Build
TS-110
Static site at the edge
Web application
Content in version control, built to static HTML by a content-first framework and served from a global edge network, with a hosted pipeline rebuilding on every commit. Fast, cheap and nearly impossible to take down, as long as the site really is static.
Framework
Astro
Content
Markdown
Hosting
Cloudflare Pages
Build pipeline
GitHub Actions
Styling
Tailwind CSS4.3
Fits · Documentation, marketing and reference sites; the moment the site needs accounts or per-user content, look at TS-010 instead.
patterns DP-400 DP-440
- BC-720.20.10Application Development
- BC-1210.10.40Serverless & Edge Runtime
TS-080
Workforce identity stack
Identity
A central identity provider issuing standards-based tokens, a directory it authenticates against, automated account provisioning into applications, and a vault for machine credentials. Every application trusts the provider instead of keeping its own passwords.
Identity provider
Keycloak26.8
Authentication protocol
OpenID Connect
Authorization protocol
OAuth 2.0
Provisioning
SCIM
Directory
OpenLDAP
Machine secrets
HashiCorp Vault2.1
Fits · Organizations standardizing sign-on across many applications; most will buy the provider as a service and keep the same shape.
patterns DP-360 DP-340 DP-350
- BC-760.20Identity & Access Management
- BC-1250.40.10Single Sign-On Integration

Stacks

Stacks

Convention-driven monolith

Cross-platform mobile

Enterprise integration layer

Event-driven microservices on Kubernetes

Machine learning platform

Open lakehouse analytics platform

Open observability stack

Platform infrastructure baseline

Server-rendered React on serverless

Serverless API on AWS

Static site at the edge

Workforce identity stack