PostgreSQL 16GA
with pgvector for embeddings
Nimbo Database
Managed databases, one hop from your GPUs.
PostgreSQL with pgvector, MySQL, Valkey, ClickHouse and Qdrant, run by Nimbo inside your VPC. Your RAG pipelines, feature stores and app data sit on the same private network as your models, with no egress between them.
Engines
The engines AI applications actually use
MySQL 8.4GA
InnoDB, read replicasValkey 8GA
Redis-compatible cache and queuesClickHousePreview
columnar analytics and log storageQdrantPreview
vector search for RAGOperations included
We run it. You query it.
High availability
Synchronous standby in a second failure domain with automatic failover. Read replicas on demand.
Backups and point-in-time restore
Daily snapshots plus continuous log archiving; restore to any second inside your retention window of 7 to 30 days.
Private by default
Databases live in your VPC with no public endpoint unless you add one and a firewall rule to match.
Encryption
Encrypted at rest with customer-managed keys available; TLS for every connection.
Upgrades on your schedule
Minor versions patched automatically in your maintenance window; major upgrades tested on a clone first.
On-premises too
Run the same managed engines on your Nimbo Rack or attached Kubernetes through Fleet, for data that must stay on site.
Put your vectors next to your models.
Create a PostgreSQL with pgvector in the same VPC as your inference endpoint and skip the egress bill.