Databases overview
Every managed database on StackBlaze can run as a single Standard instance or as a replicated, high-availability deployment. Pick the topology per database when you create it.
Standard vs high availability
Each database add-on supports two deployment topologies. You choose between them when you create the instance (and you can scale up later by raising the replica count).
Standard
A single instance with its own persistent volume. It is the simplest and lowest-cost option, and it is a good fit for development, internal tools, and workloads that can tolerate a short restart during maintenance or a node move. There is no automatic failover: if the pod or its node is rescheduled, the database is briefly unavailable until it comes back.
High availability (HA)
A clustered deployment with a primary plus one or more replicas spread across different nodes. Writes go to the primary and stream to the replicas; if the primary fails, the cluster promotes a replica automatically so the database stays available. HA also lets you serve reads from replicas to take load off the primary. It uses more resources, so it is the right choice for production and anything with an uptime requirement.
| Standard | High availability | |
|---|---|---|
| Instances | Single | Primary + 1 or more replicas |
| Node spread | One node | Replicas on separate nodes |
| Failover | None (restart) | Automatic primary promotion |
| Read scaling | Primary only | Reads can fan out to replicas |
| Best for | Dev, internal, non-critical | Production, uptime SLAs |
Choosing a topology
Available databases
StackBlaze provisions each database through a Kubernetes operator, so the same engine you run locally runs in your cluster with backups, private networking, and zero-config connection strings. The table below lists every managed database and the topologies it supports.
Relational (SQL)
| Database | Engine | Topologies |
|---|---|---|
| PostgreSQL | CloudNativePG | Standard + HA |
| MySQL | MariaDB (MySQL-compatible) | Standard + HA |
| CockroachDB | Distributed SQL | Standard + HA |
Key-value & cache
Document
| Database | Engine | Topologies |
|---|---|---|
| MongoDB | DocumentDB (MongoDB-compatible) | Standard + HA |
| CouchDB | Apache CouchDB | Standard + HA |
Search & vector
| Database | Engine | Topologies |
|---|---|---|
| OpenSearch | OpenSearch | Standard + HA |
| Milvus | Milvus (vector DB) | Standard + HA |
Analytics
| Database | Engine | Topologies |
|---|---|---|
| ClickHouse | ClickHouse | Standard + HA |
Wide-column
| Database | Engine | Topologies |
|---|---|---|
| Cassandra | Apache Cassandra | Standard + HA |
| ScyllaDB | ScyllaDB (Cassandra-compatible) | Standard + HA |
One engine, the names you know
DocumentDB (MongoDB wire protocol), Redis by Valkey, and MySQL by MariaDB. Your existing drivers, connection strings, and tools work unchanged.See Backups & recovery for the backup and point-in-time-recovery behavior that applies across every database.
Looking for messaging? Kafka and RabbitMQ are documented under Messaging.