Zoorik monitors your infrastructure without pause — shrinking and expanding block storage as demand moves, and balancing Kubernetes workloads and nodes as they run. No downtime, no maintenance window.
Cut block-storage spend up to 60–70% and end surprise I/O outages. Connect once and Zoorik right-sizes itself.
Pods ask for CPU from a number nobody revisited — then nodes get bought to satisfy it. Below: 22 CPU provisioned, 15.53 requested, carrying 3.09 CPU of real work.
Three steps, and the first two change nothing in your cluster.
Apply the agent manifest. It reads your cluster and your disks; it changes nothing. It reads six nodes and seventy-seven workloads, and writes nothing back.
One manifest, applied read-only.
Every node, workload and volume, priced. The gap between what you hold and what runs: 22 CPU held, and 3.09 CPU carrying the real work.
Every node, workload and volume, priced.
Capacity moves to match demand and keeps moving. The invoice follows; reverts are deducted, not hidden. In the modelled cluster $608.76 a month on 22 CPU becomes $194.88 on 16.
It keeps moving. Reverts are deducted, not hidden.
Disks under 2 TB are free
Read-only for the first 14 days. We measure and change nothing until you approve the number.
It is the part that should make you nervous, so it is the part with the most constraints. The filesystem is shrunk before the volume, never the other way round. A snapshot is taken first and held until verification passes. You set a headroom floor that Zoorik will not cross on any volume for any reason, and if anything looks wrong after the resize it rolls back to the snapshot automatically.
No. Volumes stay mounted and serving reads and writes throughout. That is the whole point of doing this continuously rather than in a quarterly cleanup: if a resize needed a window, it could only ever happen a few times a year, and your capacity would drift the rest of the time.
Expansion never waits on a policy window. Demand is forecast from the trend, so in most cases capacity is already there before the workload needs it. If a spike outruns the forecast, growth is immediate and unthrottled — Zoorik will always rather over-provision briefly than let a volume fill.
Zoorik works within the guardrails the cluster already declares. PodDisruptionBudgets are respected, drains are gradual, and nodes are only retired once their pods are confirmed rescheduled and healthy. If a drain cannot complete inside your budgets, it stops and reports rather than forcing it.
Block volumes on ext4 or XFS. The Elastic Engine runs on Azure; the K8s Engine supports Azure and AWS. The CSI path covers any driver that implements online expansion. Talk to us about anything you do not see listed.
On verified savings only — the difference between what you were provisioned for and what you are provisioned for now, measured against your provider's own billing data. If a month reclaims nothing, that month costs nothing. Observe stays free regardless.
Read access to metrics and volume metadata during the observe period. Write access is granted separately, per cluster, and only when you turn it on after the first 14 days. Every action, its trigger and every byte moved is recorded in an exportable audit trail.
Thirty minutes on your own fleet, in observe-only mode. You see the waste report before anything is ever written.
We reply within one business day with a scheduling link · hello@zoorik.com
