Apache Cassandra on Kubernetes: Scalable Event and Graph Systems

Scaling modern applications requires a database that can handle high write throughput, distributed storage, and predictable horizontal scaling. Apache Cassandra has become a go-to choice for companies like Netflix, which uses it as its flagship database to power massive-scale streaming, recommendation, and state-driven systems. While Netflix relies on proprietary tools for backup and monitoring, Cassandra can also integrate with open-source solutions for persistence, backup, and observability in Kubernetes environments.

---

Cassandra on Kubernetes

Running Cassandra on Kubernetes allows teams to leverage container orchestration while maintaining Cassandra’s core strengths.

Key Benefits:

• Horizontal Scalability: Cassandra scales by adding nodes, distributing partitions automatically.
• Persistence: Use StatefulSets with Persistent Volumes to ensure pods retain data across restarts.
• Backup Options:
  • Netflix uses internal tools, but open-source options like K8ssandra, Medusa, or Cassandra Reaper provide backup and repair functionality.
• High Availability: Replication across nodes and racks ensures fault tolerance.
• Operational Control: Kubernetes operators (like the CassOperator) handle rolling updates, repairs, and cluster management.

---

Example: Pods Persistence

apiVersion: apps/v1
kind: StatefulSet
metadata:
  name: cassandra
spec:
  serviceName: "cassandra"
  replicas: 3
  selector:
    matchLabels:
      app: cassandra
  template:
    metadata:
      labels:
        app: cassandra
    spec:
      containers:
      - name: cassandra
        image: cassandra:4.1
        ports:
        - containerPort: 9042
        volumeMounts:
        - name: cassandra-data
          mountPath: /var/lib/cassandra
  volumeClaimTemplates:
  - metadata:
      name: cassandra-data
    spec:
      accessModes: [ "ReadWriteOnce" ]
      resources:
        requests:
          storage: 50Gi

This ensures each Cassandra pod has persistent storage and can recover state after rescheduling.

---

Cassandra Use Cases

1️⃣ Event or State-Driven Systems

Cassandra’s design makes it ideal for:

• Time-series data
• Event logs
• User activity tracking
• IoT or blockchain events

Example Table: User Activity

CREATE TABLE user_activity (
  user_id text,
  event_date date,
  event_time timestamp,
  event_type text,
  metadata text,
  PRIMARY KEY ((user_id, event_date), event_time)
) WITH CLUSTERING ORDER BY (event_time DESC);

Query:

SELECT * FROM user_activity
WHERE user_id = 'u1'
AND event_date = '2026-02-27';

---

2️⃣ Hierarchical or Structured Data

Cassandra can also model trees or hierarchical structures using:

• Adjacency list pattern: Parent → Children
• Materialized paths: Store full path as a string
• Denormalized tree: Precomputed levels for predictable queries

Example Table: Category Hierarchy

CREATE TABLE category_children (
  parent_id text,
  child_id text,
  name text,
  PRIMARY KEY (parent_id, child_id)
);

Querying children is fast, but full tree traversal must be done at the application layer or via graph tools.

---

3️⃣ Graph Modeling with JanusGraph

For social graphs, recommendations, or complex relationships, Cassandra can serve as the storage backend for JanusGraph.

Example Tables Generated by JanusGraph:

Vertices Table

Edges Table

Gremlin Example:

g.V().has('vertex_id','u1').out('FRIEND').values('name')

• Retrieves all friends of user u1
• Supports friends-of-friends, mutuals, and weighted relationships efficiently

---

Cassandra vs CouchDB Mango Queries

Insight: Cassandra excels when high throughput, distributed writes, and predictable partitioning are required. CouchDB shines for document-based, offline-first apps with flexible querying.

---

Ideal for LangGraph-Based Systems

Cassandra’s partitioned, scalable architecture makes it suitable for LangGraph pipelines where:

• Event streams are stored as time-series tables
• Graph relationships are maintained in JanusGraph
• Real-time traversal or recommendation is needed
• Horizontal scalability is critical

Conclusion

Apache Cassandra on Kubernetes is a powerful choice for:

• Event-driven architectures
• State-driven systems
• Large-scale social graphs with JanusGraph
• Hierarchical modeling for predictable queries
• LangGraph-based AI/data pipelines

With persistent pods, backup tools, and flexible modeling patterns, Cassandra provides a scalable, distributed foundation that Netflix and other leaders trust to power their mission-critical systems.