Detailed comparison between Redis Cache and Memcached Cache

1. Quick Overview

2. Performance

• Memcached
  • Multi‑threaded, leverages multiple CPU cores for extremely high throughput on simple GET/SET.
  • Typical latency: around 0.1–0.5 ms on standard workloads.
• Redis
  • Single‑threaded but highly optimized; pipelining and I/O multiplexing can reach >100 000 ops/sec on commodity hardware.
  • Supports pipelining to reduce round trips and Lua scripting to batch operations atomically.

Head‑to‑head, for pure GET/SET workloads Memcached may edge out Redis by a few percent in throughput; however, when using pipelining, batching, or advanced data structures, Redis often outperforms.

3. Advantages & Disadvantages

Redis

Advantages

1. Rich data types: lists, sets, hashes, sorted sets, streams, etc., ideal for caching, leaderboards, queues, and more.
2. Persistence: RDB snapshots and AOF logs allow durable storage and recovery.
3. High‑availability: supports replicas, Sentinel for auto‑failover, and clustering for sharding.
4. Pub/Sub & Streams: native support for messaging and event streaming.
5. Lua scripting & transactions: atomic multi‑command operations, fewer network round‑trips.

Disadvantages

1. Single‑threaded: a single instance uses one CPU core, limiting per‑instance parallelism.
2. Memory overhead: rich data structures consume more RAM compared to Memcached.
3. Operational complexity: more features require careful configuration and monitoring.

Memcached

Advantages

1. Simplicity & lightweight: key–value only, minimal footprint, easy to deploy.
2. Multi‑threaded: naturally leverages all available CPU cores for high throughput.
3. Memory‑efficient: very low per‑item overhead, perfect for pure caching.
4. Mature ecosystem: many client libraries, plugins, and modules.

Disadvantages

1. No persistence: cache is lost on restart.
2. Limited data model: only key–value, no advanced structures.
3. No built‑in HA: sharding and failover must be managed externally.

4. Basic Deployment Steps

Deploying Memcached

Install

sudo apt-get update
sudo apt-get install memcached

Configure (/etc/memcached.conf)

• -m: memory in MB
• -p: port (default 11211)
• -u: run user
• -t: number of threads

Start & monitor

systemctl enable memcached
systemctl start memcached

Check stats via memcached-tool or telnet stats.

Sharding: phân chia key qua client hoặc proxy (Twemproxy,…).

Deploying Redis

Install

sudo apt-get update
sudo apt-get install redis-server

Configure (/etc/redis/redis.conf)

• maxmemory & maxmemory‑policy
• Enable persistence: save (RDB), appendonly yes (AOF)
• Replication: replicaof host port
• Clustering options if needed

Start & monitor

systemctl enable redis-server 

systemctl start redis-server

Use redis-cli info to inspect stats, replication, and memory usage.

1. Scaling
   • Replica: add Redis replicas for read scaling.
   • Sentinel: configure for auto‑failover HA.
   • Cluster mode: automatic sharding across multiple nodes.

5. When to Choose Which

Conclusion

• Use Memcached when your application demands a simple, ultra‑fast in‑memory cache with minimal operational overhead and no need for persistence or complex data types.
• Use Redis when you require rich data structures, durability, replication/HA, pub/sub, or scripting capabilities, or when you plan to scale out via clustering.