What should I use for shared cache across app servers?

For a shared cache across app servers, the usual choice is a distributed cache:

Good options

• Redis — most common choice
  • Fast
  • Widely supported
  • Good for session caching, computed results, rate limiting, pub/sub, locks
• Memcached
  • Very simple and very fast
  • Best for pure ephemeral key/value caching
  • Less feature-rich than Redis
• Managed cloud cache
  • AWS ElastiCache (Redis/Memcached)
  • Azure Cache for Redis
  • Google Cloud Memorystore

What I’d recommend

• Use Redis if you want the safest default.
• Use Memcached only if you want a very simple cache and don’t need Redis features.

If you need stronger guarantees

If the data must be:

• durable,
• strongly consistent,
• or shared as source-of-truth,

then a cache is probably the wrong tool — consider a database or another shared state store.

Rule of thumb

• Local in-process cache: fastest, but per-server only
• Distributed cache (Redis/Memcached): shared across app servers
• Database: for durable shared data, not just caching

If you want, I can recommend the best choice based on your stack (language, traffic, cloud, and whether you need TTL, pub/sub, or session storage).