What should I use instead of a database for hot reads?

If you mean “what can handle very fast, frequent reads without hitting the primary database every time?” the usual answer is:

Use a cache or read-optimized store, depending on the need

1) In-memory cache

Best when you want the fastest possible hot reads.

Examples:

• Redis
• Memcached
• KeyDB
• local in-process cache like Caffeine (Java), Guava, or LRU caches

Use this when:

• data is read very often
• staleness of a few seconds/minutes is acceptable
• you want to reduce load on your main DB

2) CDN / edge cache

Best for:

• static content
• public content
• API responses that can be cached near users

Examples:

• Cloudflare
• Fastly
• Akamai
• AWS CloudFront

3) Search/index store

If the “hot reads” are lookup/search-heavy rather than transactional.

Examples:

• Elasticsearch
• OpenSearch
• Solr

4) Read replicas

If you still need a database but want to offload reads:

• PostgreSQL/MySQL read replicas
• separate reporting/serving database

5) Materialized views / denormalized read models

If reads are predictable and can be precomputed:

• materialized views
• CQRS read models
• precomputed aggregates in Redis or another store

---

Rule of thumb

• Fastest hot reads: Redis / in-memory cache
• Global low-latency delivery: CDN/edge cache
• Search/filter-heavy reads: Elasticsearch/OpenSearch
• Still need SQL consistency: read replicas or materialized views

If you tell me:

1. what kind of data you’re reading,
2. how fresh it must be,
3. expected traffic, I can suggest the best option more precisely.