Moving billions of messages from MongoDB to Cassandra
How Discord escaped MongoDB tail-latency degradation by rethinking their message data model around Cassandra partition semantics.
Deadline driven preparation
- One path, in order
- Your code and progress stay in this browser
- Pick a stage and start
Machine coding
48 lessons
Language Fundamentals
Data Structures
Algorithms
48 lessonsStart
Production engineering
45 lessons
Frontend
Backend
DevOps
45 lessonsStart
System design
38 lessons
Core Concepts
Low Level Design
High Level Design
38 lessonsStart
Last Mile Prep
10 lessons
HM Round
AI Round
Triage & Telemetry
10 lessonsStart
Technologies
GitTrack an AI wrapper from local setup through pull requests, recovery, and releases.
DockerPackage an application and its dependencies into an image you can build, run, and share.
KubernetesKeep container workloads running as instances fail, traffic grows, and versions change.
AWS CLITurn cloud-console operations into repeatable API queries and scripts.
TerraformDescribe infrastructure in code, review a plan, and track the resources it manages.
HelmPackage Kubernetes resources into reusable charts with environment-specific values.
Linux and networkingTrace a service failure through processes, resource limits, DNS, TLS, and HTTP.
SnowflakeGive analytical queries their own compute while sharing a managed data store.
PostgresUse relational constraints, transactions, and query plans for concurrent application data.
RedisShare cached values and operate on in-memory data structures across application instances.
KafkaPublish retained events so independent consumers can process and replay them.
CassandraDistribute high-volume records around known queries, partition keys, and tunable consistency.
ElasticsearchTurn documents into searchable terms, ranked results, and filtered aggregations.
BookShelf
Frontend Engineering
Browser rendering pipelines, reactive state machines, hydration, accessibility, and production performance.
Rendering architecture
State & reactivity
Production performance
Backend Engineering
Request execution, concurrency boundaries, storage engines, transactions, and event-driven message queues.
Requests & protocols
Storage & transactions
Queues & replication
DevOps
Containerization, Kubernetes cluster orchestration, ingress routing, CI/CD, and zero-downtime rollouts.
Containers & processes
Kubernetes orchestration
CI/CD & release pipelines
High-Level Design (HLD)
15 end-to-end distributed system blueprints with request flows, database schemas, and interactive Excalidraw drawings.
Storage & caching tiers
Streaming & real-time chat
Global services & search
Low-Level Design (LLD)
15 machine coding problems emphasizing extensible OOP patterns, concurrency control, and test-driven domain modeling.
Resource allocators (Parking lot)
Caches & rate limiters
Databases & schedulers
Language Fundamentals
Deep language runtime semantics, memory models, garbage collection, and idioms across Python, Go, JS, Java, and C++.
Python runtime & idioms
Go, Java & C++ toolchains
JavaScript V8 & async loops
Case Studies
From fanout-on-write to hybrid delivery
How Twitter's news feed moved away from pre-computing timelines for all followers as accounts reached millions of followers.
The BGP leak that took down Cloudflare
How a single misconfigured route advertisement propagated globally and what Cloudflare changed about their route filtering and BGP safety nets.
The Slack message storm incident
How a feedback loop between presence updates and reconnecting clients took down Slack and what changes in their notification and connection handling prevented recurrence.