Managed Redis/Valkey (Deferred)
Deferred
Redis/Valkey is evaluated but not on the current committed roadmap. This document captures the design direction for when the service is prioritized. See Portfolio Strategy for sequencing criteria.
Purpose: For platform engineers, documents the design direction for a managed Redis/Valkey caching service — capabilities under consideration, deployment model, and what would trigger roadmap inclusion.
Rationale for Deferral
Per the portfolio strategy, a data service enters the roadmap when it meets all selection criteria:
- ✅ Operator maturity — Redis operators exist (redis-operator, Spotahome, DragonflyDB)
- ⚠️ Customer demand — acknowledged but not yet at threshold for prioritization
- ⚠️ Operational clarity — day-2 operations (failover, persistence, scaling) need validation
- ✅ Air-gap viability — standard container images
- ⚠️ Commercial clarity — revenue model not yet defined
Design Direction (When Prioritized)
Capabilities Under Consideration
| Capability | Description |
|---|---|
| Sentinel mode | Automated failover with Redis Sentinel |
| Cluster mode | Sharded data across multiple nodes |
| Persistence | RDB snapshots + AOF for durability options |
| TLS | cert-manager issued certificates |
| Authentication | Password-based + ACL (Redis 6+) |
| Monitoring | Prometheus exporter + Grafana dashboards |
| GitOps lifecycle | CRDs in Git, FluxCD reconciliation |
| Memory policies | Configurable eviction (allkeys-lru, volatile-lru, noeviction, etc.) |
Service Tiers (Draft)
| Tier | Topology | Persistence | Use Case |
|---|---|---|---|
| Development | Single node, no persistence | None | Local development, testing |
| Standard | Sentinel (1 primary + 2 replicas) | RDB snapshots | Production caching |
| Premium | Redis Cluster (3+ shards) | RDB + AOF | High-throughput, large datasets |
Operator Candidates
| Operator | Status | Notes |
|---|---|---|
| Spotahome redis-operator | Maintained | Sentinel mode, basic lifecycle |
| OpsTree Redis Operator | Active | Cluster + Sentinel, Prometheus |
| DragonflyDB (Redis-compatible) | Active | Alternative engine, higher performance claims |
No operator selection has been made. Evaluation will start when the service enters the roadmap.
Integration Points
| Integration | Mechanism |
|---|---|
| Platform observability | Prometheus exporter → kube-prometheus-stack |
| Security | NetworkPolicies + TLS + authentication |
| Backup | Velero PVC snapshots (for persistent modes) |
| GitOps | FluxCD reconciliation of Redis CRDs |
What Would Trigger Prioritization
- Multiple customer requests with production urgency
- Streaming or PostgreSQL services creating caching dependency
- Clear commercial model (tier pricing, support boundaries)
- Operator maturity validated in production context
Further Reading
- Data Services Overview — family overview and principles
- Portfolio Strategy — selection criteria and roadmap