kvgo

Sharded in-memory key-value store in Go. Custom TCP server, per-shard RWMutex locking, RESP2 protocol, Prometheus/Grafana observability.

Architecture

client (TCP)
  ||
  \/
server.Listen -> goroutine-per-connection
  ||
  \/
command parser -> command router (GET, SET, DEL)
  ||
  \/
ShardedStore (N shards, FNV-32a hash)
  |-> shard[0] { sync.RWMutex, map[string]string }
  |-> shard[1] { ... }
  |-> shard[N-1] { ... }

Keys are distributed across runtime.NumCPU() shards via FNV-32a so concurrent reads on different shards never contend. Prometheus metrics (latency histograms, hit/miss ratio, active connections, shard distribution) are exposed on :2112/metrics with a provisioned Grafana dashboard.

Commands

Command	Description
SET key value	Store a key-value pair
GET key	Retrieve a value by key
DEL key	Delete a key
PING	Test server liveness
COMMAND COUNT	Return the number of supported commands

Quick start

make docker-up       # kvgo + Prometheus + Grafana
make run             # local binary only
make test            # go test -v -race ./...

redis-cli -p 6379 SET foo bar
redis-cli -p 6379 GET foo

• kvgo: localhost:6379
• Prometheus: localhost:9090
• Grafana: localhost:3000 (dashboard auto-provisioned)

Roadmap

• Worker pool: bounded goroutine pool to cap resource usage under high connection counts.
• Raft consensus: multi-node replication with leader election for fault tolerance.