
VEXXHOST at OpenInfra Day North America 2026
VEXXHOST recaps OpenInfra Day North America 2026 in Chicago, highlighting its headline sponsorship, Mohammed Naser's talks, and the future of OpenStack operations.
Read field noteField notes / Latest
Engineering notes from operating open infrastructure: the failures, design decisions, and upstream work that make open infrastructure better.
Browse all field notes
VEXXHOST recaps OpenInfra Day North America 2026 in Chicago, highlighting its headline sponsorship, Mohammed Naser's talks, and the future of OpenStack operations.
Read field note
Ceph vs SAN: Compare performance, scalability, licensing, and operational costs to choose the right enterprise storage modernization strategy.
Read field note
Cloud exit strategy checklist: Review data ownership, exit costs, vendor lock-in, and migration support before signing a cloud contract.
Read field noteThe exit terms to settle before signing a managed private cloud contract: data and configuration export, transition assistance, runbooks, hardware and IP ownership, notice periods and fees, with what DORA and the EU Data Act require.
A managed private cloud contract usually runs for years, and nobody wants to argue about the exit clause at signing. A managed private cloud is a dedicated cloud that a provider deploys and operates for you, in its data centre or in yours. Exit terms are the clauses that decide how you leave it: what you can take, how long the provider keeps running things while you move, and what leaving costs.
Two EU rules made those clauses a compliance matter for many buyers. The EU's Digital Operational Resilience Act (DORA) has applied since January 17, 2025. It requires financial entities to hold exit strategies for ICT services that support critical or important functions. The EU Data Act's switching rules for data processing services have applied since September 12, 2025.
A weak exit clause stays invisible until the day you use it. On that day you may face a notice period far shorter than the migration it has to cover, a transition priced at whatever the incumbent quotes, or a configuration only the provider's staff can read. This memo takes each exit term, says what it protects in money and time, and ends with a rule that sorts the terms into must-have and negotiable.
DORA's exit requirement has a defined scope. Article 28(8) applies to financial entities, and only to ICT services supporting critical or important functions. For those services, the entity must:
Article 30 turns that into contract language. Contracts for critical or important functions must include an exit strategy with a mandatory adequate transition period, during which the provider keeps delivering the service while the entity migrates. Every ICT contract with a financial entity must cover the return of data in an easily accessible format if the contract ends or the provider fails, and set termination rights with minimum notice periods.
Banks and other institutions under the EBA's outsourcing rules had a head start. The EBA's 2019 outsourcing guidelines already asked for documented exit plans for critical or important outsourced functions, tested where appropriate.
The Data Act reaches further than finance. It covers data processing services, and it applies to providers serving customers in the EU wherever the provider is established. Its switching terms set contract rules for covered services:
Some services are exempt from the Act's rules on switching charges and from some of its technical obligations. The exemption covers services with most of their main features custom-built for one customer, or with all components developed for that customer, and not offered at broad commercial scale. The contract terms above still apply to them. Dedicated hardware on its own does not meet that test. Whether a managed private cloud qualifies depends on how it is built and sold, so write the terms into the contract either way. Your counsel decides how each rule applies to your agreement; this memo covers the commercial side.
Data portability needs more than a promise of "your data", which leaves out most of what you need to run the workloads somewhere else. On an open source stack of OpenStack, Kubernetes and Ceph, a starting inventory for the export has named contents and open formats:
The contract should list these by name, state the format of each and set who pays for the export work. Ask for the team that would receive the workload to test a restore of one representative workload during the first year. You learn whether the list holds while the relationship is good.
Open source changes what the configuration export is worth. Atmosphere, the OpenStack platform we build and run, is open source, with its code publicly available on GitHub. For another team to redeploy your cloud with the same public code, the handover needs your deployment configuration, its secrets and the keys to decrypt them, the Atmosphere version in use, and the network and hardware specifications.
Transition assistance is the period after notice when the outgoing provider keeps running production and helps move it. Exit disputes tend to start here, when both sides find out what was never written down.
Size the transition from your own migration estimate, planned in waves and counting testing and change windows.
A managed provider runs your cloud through automation and runbooks: deployment playbooks, upgrade procedures, alert responses and the notes behind each exception. Without them, a new operator inherits a working cloud and no record of how it is kept working.
Ask for three things in the contract:
On Atmosphere, that statement can be short, because its deployment playbooks and upgrade procedure are public in the repository.
Ownership decides whether an exit is a migration or a renumbering project on top of one.
Atmosphere comes in hosted and on-premise editions. The on-premise edition runs in your data centre on hardware you choose, so a team that wants to own its hardware can start with it.
Addresses assigned from a provider's block normally stay with that provider. RIPE NCC policy is explicit that provider aggregatable addresses have to be returned and the network renumbered when you change provider. Provider independent space can move with you under that policy, if the new provider agrees to announce it. If customers, partners or firewalls allow-list your public addresses, either bring address space you already hold and have the provider announce it, or write a renumbering window into the transition.
Three lines set the price of leaving, and they interact.
Keep termination for cause outside all three. DORA Article 28(7) lists the cases where a financial entity must be able to terminate, including a significant breach by the provider. We recommend that every buyer's contract keeps fees off those cases.
Exit terms cost money at signing. A provider that commits to a long transition at fixed rates, named staff and a tested export is reserving people and pricing that risk into the contract. A buyout option on hardware also changes how the provider finances that hardware.
For a small team, standard terms can be the right trade. That holds when:
The take stops holding when:
Sort the exit schedule of every managed private cloud contract into three tiers, and check what moves an item between them:
With Atmosphere, the deployment and upgrade code those terms depend on is already public on GitHub, and the code that runs a customer's cloud is open for any operator to deploy. We have contributed to OpenStack since the Bexar release in 2011, and we follow a policy to contribute every fix back upstream. Deployment and migration support sit with our professional services team.
Before you return the redline on an exit schedule, you can talk through its terms with our team.
Choose from Atmosphere Cloud, Hosted, or On-Premise.
Simplify your cloud operations with our intuitive dashboard.
Run it yourself, tap our expert support, or opt for full remote operations.
Leverage Terraform, Ansible or APIs directly powered by OpenStack & Kubernetes