VALKEY // DATABASES
Valkey on Your Cluster
Run Valkey for caches, queues, or durable key-value storage on your own cluster. Configure persistence, back up to your object storage, restore snapshots, and enable TLS and Sentinel failover.
Powered by Valkey
Available versions
FEATURES // VALKEY
Valkey Storage and Operations
Edka provisions Valkey on your own cluster and gives you a clean control surface for day-2 operations. The data stays in your cloud account.
Topology & Resources
Size the runtime for your workload.
- Standalone or primary with up to three replicas
- Separate CPU and memory requests and limits
- Node-pool placement and taint tolerations
Persistence & Storage
Choose ephemeral caching or persistent data.
- Durable defaults: RDB + AOF, flushed every second
- Configurable dataset memory budget
- Guarded persistence changes and volume expansion
Automatic Failover
Opt in to promotion through Valkey Sentinel.
- Three Sentinels with quorum two
- Requires ACL auth and at least three data nodes on separate workers
- Stable application endpoint and optional read service
Backups & Restore
Run backup and restore jobs inside your cluster.
- Scheduled and manual RDB snapshots to S3-compatible storage
- Snapshot restore into a new installation
- Checksum validation and fresh credentials on restore
Access & TLS
Require authentication and TLS for public access.
- Managed certificates with renewal and CA download
- Public LoadBalancer, private MetalLB, or Tailscale exposure
- Existing internal application service retained
Insights & Alerts
Inspect the primary, replicas, and Sentinel readiness.
- Logical key counts from the observed primary
- Per-instance rates and replica synchronization health
- Resource metrics, logs, and an editable Valkey alert pack
Other engines
PostgreSQL 
MySQL 
ClickHouse GET STARTED // NOW
Provision Valkey
Run Valkey and its backup jobs on infrastructure you control.
Which Valkey versions can I deploy with Edka?
Available versions today: 9.1.2, 9.0.2. You pick the version when provisioning the runtime.
Is Valkey on Edka a cache or a durable store?
Either. Choose no persistence, RDB, or RDB with AOF. New durable installations default to RDB + AOF with a flush every second and a 50% dataset memory budget. Persistent volumes can grow when the storage class supports expansion; shrinking or removing them is blocked.
Does Valkey support backups and restore?
Yes. Scheduled and manual jobs save RDB snapshots to your S3-compatible storage. Restore creates a new installation at the recorded Valkey version with fresh credentials. Restore checks the snapshot checksum before loading it. Restore includes the selected snapshot only; it does not recover writes made afterward or provide point-in-time recovery.
What alerts are available for Valkey?
Install the Valkey pack from Observability → Alerts → Packs. Ten editable rules cover exporter and server availability, memory and connection pressure, rejected connections, replica connectivity and traffic, RDB and AOF failures, and missing core metrics. It uses the existing Valkey exporter and requires VictoriaMetrics.
What does automatic failover require?
Enable Sentinel failover with ACL authentication, a primary and at least two replicas, and a separate schedulable worker for each data instance. Three Sentinels elect a primary with quorum two. Applications keep the same service endpoint and must reconnect after failover. Replication is asynchronous, so acknowledged writes can still be lost during failures. Removing failover requires restoring into a new installation.
How is Valkey access secured?
Public LoadBalancer access requires ACL authentication and managed TLS. Download the CA and configure your client to verify both the certificate and server name. Private MetalLB supports source CIDR restrictions; public Hetzner load balancers do not. Tailscale uses tailnet access rules. The existing internal service remains available without TLS for managed applications; password or port changes require reapplying those applications.
Where does my Valkey data live?
In your own Kubernetes cluster, on your own infrastructure. There is no hosted database service in between. You keep full control of the runtime and its storage.