https://github.com/michelderu/cassandra-fundamentals
Cassandra training and lab material
https://github.com/michelderu/cassandra-fundamentals
cassandra-database
Last synced: 2 months ago
JSON representation
Cassandra training and lab material
- Host: GitHub
- URL: https://github.com/michelderu/cassandra-fundamentals
- Owner: michelderu
- Created: 2026-03-30T12:20:18.000Z (3 months ago)
- Default Branch: main
- Last Pushed: 2026-03-30T13:51:40.000Z (3 months ago)
- Last Synced: 2026-03-30T15:11:31.863Z (3 months ago)
- Topics: cassandra-database
- Homepage:
- Size: 56.9 MB
- Stars: 0
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Cassandra training — architecture and data modeling
## About Apache Cassandra
[Apache Cassandra](https://cassandra.apache.org/) is an open source, **distributed wide-column** database designed for **massive scale**, **high availability**, and **predictable low latency** on commodity hardware or in the cloud. It uses a **masterless**, peer-to-peer topology: every node can serve reads and writes, and data is replicated across the cluster with **tunable consistency** so applications can trade latency against how many replicas must agree on each operation.
People use Cassandra as an **operational data store** for live workloads—time series and metrics, event logging, product catalogs, session and profile data, messaging back ends, IoT ingestion, and increasingly **AI/ML** and retrieval-style pipelines where throughput and uptime matter more than ad-hoc relational joins. The project describes it as trusted by **thousands of companies** with large active data sets; release testing includes clusters of up to **1,000 nodes**. A public case study on the Cassandra site quotes **Bloomberg** serving **more than 20 billion requests per day** on a **~1 PB** dataset across **1,700+** nodes. The **2024 Apache Cassandra user survey** published **140** responses on use cases, deployment size, and experience. See [References](#references) for links.
## This repository
This repo is **hands-on training** in two parts:
1. **Architecture** — You run a **three-node** cluster (Docker Compose) and work through **internals and operations** in [`architecture/`](architecture/README.md): placement, consistency, gossip, the storage engine, and repairs / LWT.
2. **Data modeling** — A **seven-module** track in [`data-modeling/`](data-modeling/README.md) teaches **query-first** schema design: partition keys, clustering, denormalization, and anti-patterns. It includes **hands-on labs in each module** on the **same Docker Compose cluster** ([`docker-compose.yml`](docker-compose.yml)). Create `lab_ks` / `events` per [architecture/02-lab-environment.md](architecture/02-lab-environment.md) before module **02**.
You can complete **architecture** first, then **data modeling**, or jump to data modeling if you already run Cassandra—still use Compose and [module 02](architecture/02-lab-environment.md) for the shared schema before the hands-on exercises.
## Learning path
### Architecture (cluster labs)
| Module | File |
|--------|------|
| 01 — Architecture and deployment | [01-architecture-and-deployment.md](architecture/01-architecture-and-deployment.md) |
| 02 — Lab environment | [02-lab-environment.md](architecture/02-lab-environment.md) |
| 03 — Masterless, peers, placement | [03-masterless-peers-and-placement.md](architecture/03-masterless-peers-and-placement.md) |
| 04 — CAP and tunable consistency | [04-cap-and-tunable-consistency.md](architecture/04-cap-and-tunable-consistency.md) |
| 05 — Gossip and topology | [05-gossip-and-topology.md](architecture/05-gossip-and-topology.md) |
| 06 — Storage engine (write/read, compaction, tombstones) | [06-storage-engine-write-through-read.md](architecture/06-storage-engine-write-through-read.md) |
| 07 — Self-healing, LWT, summary | [07-self-healing-lwt-and-summary.md](architecture/07-self-healing-lwt-and-summary.md) |
### Data modeling (CQL labs)
| Module | File |
|--------|------|
| 01 — Intro and paradigm | [01-intro-and-paradigm.md](data-modeling/01-intro-and-paradigm.md) |
| 02 — Process and primary key | [02-process-and-primary-key.md](data-modeling/02-process-and-primary-key.md) |
| 03 — Placement and partition health | [03-placement-and-partition-health.md](data-modeling/03-placement-and-partition-health.md) |
| 04 — Clustering and wide partitions | [04-clustering-and-wide-partitions.md](data-modeling/04-clustering-and-wide-partitions.md) |
| 05 — Tombstones and denormalization | [05-tombstones-and-denormalization.md](data-modeling/05-tombstones-and-denormalization.md) |
| 06 — Anti-patterns | [06-anti-patterns.md](data-modeling/06-anti-patterns.md) |
| 07 — Checklist, labs, blueprint | [07-checklist-labs-and-blueprint.md](data-modeling/07-checklist-labs-and-blueprint.md) |
## Prerequisites
- Docker Desktop or Docker Engine **with Compose v2**
- About **4 GB** free RAM for the stack (heap capped at 512 MB per node in `docker-compose.yml`)
## Start the lab cluster
```bash
docker compose up -d
```
If your installation only provides Compose v1:
```bash
docker-compose up -d
```
Wait until all nodes show **UN** (up/normal):
```bash
docker exec cassandra-1 nodetool status
```
Connect with **cqlsh** (from any node):
```bash
docker exec -it cassandra-1 cqlsh cassandra-1 9042
```
The host maps **port 9042** to `cassandra-1` for drivers connecting from your machine (e.g. `127.0.0.1:9042`).
## Stop and reset
```bash
docker compose down
```
To wipe data volumes and start clean:
```bash
docker compose down -v
```
## References
1. Apache Software Foundation, *Apache Cassandra* (homepage: scale, testing, and user quotes). [https://cassandra.apache.org/](https://cassandra.apache.org/)
2. Apache Cassandra community, *2024 User Survey Results* (October 2024, n=140). [https://cassandra.apache.org/_/blog/2024-User-Survey.html](https://cassandra.apache.org/_/blog/2024-User-Survey.html)
Thanks to **David Leconte** for the architecture images used in the Architecture modules.