Backend System Design

Key Technologies

The tools that power modern distributed systems. Know not just what they are, but when to use them, what problems they solve, and what their trade-offs are.

Redis

Redis

In-Memory Data Store

Primary choice for caching, session storage, pub/sub, and distributed locks. Supports rich data structures (sorted sets, streams) making it versatile beyond simple key-value.

CachingPub/SubDistributed LocksLeaderboardsRate Limiting
Kafka

Kafka

Distributed Event Streaming

Durable, high-throughput event bus. Enables decoupled async communication, event sourcing, and real-time stream processing. Consumers can replay events from any offset.

Event StreamingAsync DecouplingReplayHigh Throughput
Elasticsearch

Elasticsearch

Distributed Search & Analytics

Full-text search, fuzzy matching, geo-search, and log analytics at scale. Uses inverted indexes. Common for search-as-a-service alongside a primary DB.

Full-text SearchGeo SearchLog AnalyticsAggregations
Cassandra

Cassandra

Wide-Column NoSQL DB

Optimized for high write throughput, linear scalability, and multi-region deployments. Tunable consistency. Ideal for time-series data, activity logs, and messaging.

Write-heavyMulti-regionTime-seriesTunable Consistency
DynamoDB

DynamoDB

Managed NoSQL (AWS)

Serverless key-value / document store with single-digit millisecond latency at any scale. Ideal for predictable access patterns. Global tables for multi-region.

Serverless ScaleLow LatencyGlobal TablesTTL
PostgreSQL

PostgreSQL

Relational Database

Battle-tested RDBMS with strong ACID guarantees. Supports JSONB, full-text search, and PostGIS for geo. Default choice when you need strong consistency and complex queries.

ACIDComplex QueriesJSONBPostGIS
Flink

Flink

Stream & Batch Processing

Stateful stream processing with exactly-once semantics. Used for real-time aggregations, event-time windowing, and complex event processing on Kafka streams.

Stream ProcessingWindowingExactly-onceStateful
ZooKeeper

ZooKeeper

Distributed Coordination

Leader election, distributed configuration, service discovery, and distributed locks. Acts as the source of truth for cluster coordination in systems like Kafka, HBase.

Leader ElectionService DiscoveryConfig MgmtDistributed Locks
API Gateway

API Gateway

Edge Routing Layer

Single entry point for all clients. Handles routing, auth, rate limiting, SSL termination, and request transformation. Decouples clients from microservice topology changes.

RoutingAuthRate LimitingSSL Termination