MayaScale · NVMe-oF block · CSI for Kubernetes

Sub-millisecond block storage, in your cloud.

MayaScale is NVMe-oF block storage for your own cloud account — volumes served from ZFS pools or LVM volume groups at sub-millisecond latency, with filesystem volumes alongside. A CSI driver brings both to any Kubernetes — k3s, EKS, GKE, AKS — so databases, vector indexes, and stateful apps run unchanged.

See it on your cloud → Read the docs one terraform deploy · AWS · GCP · Azure
Block and file

One system for everything stateful.

Sub-millisecond latency

NVMe-oF over standard TCP networking — no special hardware, no client agents. Millions of IOPS per node for transaction-heavy databases and high-concurrency vector search.

Kubernetes-native

The CSI driver provisions block and filesystem volumes on any Kubernetes — k3s, EKS, GKE, AKS — with snapshots, clones, and resize. Stateful workloads get real block performance.

Instant copy-on-write clones

Volumes clone in seconds regardless of size, storing only changes — the foundation for test databases, staging environments, and Zettabranch's Postgres branching.

Native object tier

Cold volumes and idle branches settle to object storage automatically and return on demand — no lifecycle rules, no archival jobs, nothing deleted.

Always available

Active-active storage pairs with mirrored writes and automatic failover. Applications reconnect and continue — no manual recovery.

Guardrails per volume

Storage quotas and performance tiers enforced by the platform itself — a noisy workload can't starve the one that matters.

Key capabilities

What's in the box.

NVMe-oF over TCP
Standard NVMe initiators on ordinary cloud networking — full performance from stock Linux, no client software.
Block & file volumes
Raw NVMe block devices or filesystem volumes, provisioned from ZFS pools and LVM volume groups — one system for both.
CSI driver
Dynamic provisioning, volume snapshots, clones, and resize on any conformant Kubernetes — k3s, EKS, GKE, AKS, or self-hosted.
Copy-on-write engine
Instant snapshots and clones that store only changes; checksummed, compressed, and thin-provisioned throughout.
Object storage tier
Cold data moves to S3, GCS, or Blob as live storage — carried by the engine itself, block by block, and attached back on demand.
High availability
Active-active server pairs, mirrored writes, floating endpoints, automatic failover and failback.
Quality of service
Per-volume performance tiers and quotas enforced at the storage target, below anything an application can bypass.
Deployment
Terraform from the AWS, Google Cloud, or Azure marketplace into your own account and VPC; sized configurations ready in minutes.
Postgres branching
MayaScale is the storage engine under Zettabranch — production-like database branches for every developer and agent.
Self-serve

Run it yourself.

Deploy with terraform into your own account and VPC — sized configurations, ready in minutes, billed through the cloud agreement you already have. Prefer a guided first run? Talk to us.

AWS

Terraform deployment into your AWS account — S3 as the object tier.

Google Cloud

Terraform deployment into your GCP project — GCS as the object tier.

Azure

Terraform deployment into your Azure subscription — Blob as the object tier.

zettalane-terraform — the modules, sized configurations, and step-by-step guides for every cloud, in one repository.

Block storage that keeps up with your serving tier.

Self-serve with terraform from your cloud marketplace — ready for your most demanding workloads.