Off-Clouding
Leave the public cloud without leaving its conveniences
Public cloud made it easy to start, and that was the point. The bills, however, grow together with your success, and at a certain scale renting compute by the hour becomes the most expensive way to run stable workloads.
Off-clouding is our service for moving mature workloads onto infrastructure you own. We plan the migration, build the platform, move the data and operate the result together with your team. The tools stay familiar. The invoices get smaller.
The economics favor ownership
Flexera reports that 27% of cloud spend is wasted and budgets are exceeded by 17% on average. The exits are well documented: 37signals cut a 3.2M USD annual AWS bill to under 1M USD on owned hardware and projects more than 10M USD in savings over five years, while Dropbox reduced operating costs by 74.6M USD in the two years around its move onto its own infrastructure.
Bare metal performs
Hypervisors and noisy neighbors cost real-world workloads 10 to 25% of their performance, even when benchmark-perfect environments show less. Bare metal returns every CPU cycle and every IOPS to your application, and you gain full control over hardware, firmware and tuning.
Open source keeps you compatible
Mature platforms such as OpenStack, Proxmox VE, Kubernetes, Ceph and MinIO expose APIs compatible with the major clouds, including S3. Applications move with configuration changes instead of rewrites, and your team keeps the tools and skills it already has.
The industry is already moving
A 2024 Barclays CIO survey found 83% of enterprises planning to move at least some workloads back from the public cloud, up from 43% in 2020. IDC reports about 80% expect some repatriation within a year. Andreessen Horowitz estimates repatriation can roughly halve the cloud bill for steady workloads.
What is included
Assessment and cost modeling
We start from your real bills and usage exports, then model what the same workloads cost on owned hardware, including power, space, connectivity and our operation. You see the numbers before you commit to anything.
Platform build
Compute on Proxmox VE or OpenStack, S3 compatible storage on Ceph or MinIO, Kubernetes where it fits. Built in our facilities or yours, on hardware sized for your measured load.
Staged migration
Compatible APIs mean applications move with configuration changes instead of rewrites. We migrate in stages, with rollback points at every step and the public cloud kept warm until you decide to switch it off.
Day-2 operation
After the migration, our engineers monitor and operate the platform around the clock. The field engineers who racked your hardware are the ones on call when it needs attention.
Example of use
A SaaS provider spending about 80k USD per month on public cloud moved its steady-state workloads onto two racks of owned hardware in our Osaka and Tokyo facilities, keeping burst capacity in the cloud.
The migration ran over one quarter with no customer-visible downtime. S3 compatible storage and Kubernetes meant the application code itself did not change.
- Steady-state infrastructure spend cut by more than half
- The hardware paid for itself within the first year
- Performance headroom roughly doubled on the same budget
What to expect
- Run cost for steady workloads reduced in the 30 to 70% range reported by published cloud exits
- Full hardware performance with no noisy neighbors
- No new lock-in: the platform is open source your team could run