9 Proven Techniques to Supercharge Service Performance

1. Caching

Cache is essential for performance; browsers can use Expires, Cache‑Control, Last‑Modified, and Etag headers, while services often employ Redis for in‑memory caching and MySQL's buffer pool for page caching. Choosing between Redis (consistent across machines) and local memory depends on consistency vs. speed trade‑offs. Common cache pitfalls include cache avalanche, penetration, and breakdown, each mitigated by random TTLs, Bloom filters, or distributed locks. Eviction strategies such as LRU, LFU, or random replacement are used, with Redis and MySQL both providing configurable policies. Example LRU implementation:

type LRUCache struct {
    sync.Mutex
    size int
    capacity int
    cache map[int]*DLinkNode
    head, tail *DLinkNode
}

type DLinkNode struct {
    key, value int
    pre, next *DLinkNode
}

Redis objects also store LRU/LFU metadata: <code>typedef struct redisObject { unsigned type:4; unsigned encoding:4; unsigned lru:LRU_BITS; /* LRU time or LFU data */ int refcount; void *ptr; } obj; </code>