By Michał Puchała · 2026-11-17 · 8 min read
Managed Databases in Your Own Cloud Account: How Much Control Do You Gain?
Running a managed database in your own cloud account can give you more control over infrastructure while keeping specialist support. We examine what changes, what still depends on the operator, and how to check European hosting, access, recovery and your ability to leave.

Your board wants more control over company data. Your IT team wants someone else to handle database maintenance. Running a managed database in your own cloud account offers a way to pursue both, provided you understand which responsibilities move with the infrastructure and which stay with the operator.
The model is usually called bring your own cloud, or BYOC. A database specialist operates software inside an account your company controls, using permissions you grant. Altinity's deployment guide describes this arrangement for ClickHouse, a database used for analysing large amounts of data.
The question has fresh relevance after EDB announced its acquisition of Altinity on 5 November. EDB's plans include deployment in customers' own clouds and on their own infrastructure, while the announcement distinguishes the integration roadmap from capabilities available today. Altinity separately says existing customer contracts, service commitments and products remain unchanged.
For a European business assessing this model, the useful question is what you could inspect, change, recover or take over yourself. This article explains how to evaluate those choices before committing a business-critical database.
What changes when the cloud account is yours
Account ownership gives you a direct relationship with the infrastructure beneath the database. In Altinity's model, the computing and storage resources sit in your account, and you grant Altinity access to provision and manage them. Its service introduction distinguishes that option from running in Altinity's account or supplying a Kubernetes environment, the system used to coordinate the software running across servers.
That separation can be valuable when your company wants to retain infrastructure while changing who operates it. It also gives your IT team a place to inspect the resources supporting the service. We would make that visibility part of the acceptance test: your administrators should be able to identify the database servers, storage, network connections and access permissions without asking the operator to produce every answer.
The management system may still sit elsewhere. ClickHouse Cloud's BYOC architecture, for example, keeps its management services in ClickHouse's environment while the database infrastructure runs in the customer's account. Those management services need permissions to create and maintain resources inside that account.
Engineers call this a separation between the data plane, where the database runs, and the control plane, which manages it. For a board, the practical distinction is simpler: owning the account does not establish who can perform every administrative action. Ask the operator to show which actions remain available if its management service is temporarily unreachable.
European hosting is a separate decision
BYOC describes a deployment arrangement, not a promise that a service supports every cloud. ClickHouse Cloud's overview lists AWS, Google Cloud and Azure as supported providers. Altinity also documents deployment inside a customer's Hetzner account, including creating a dedicated project and granting a read/write access token.
For a company planning to move infrastructure to a European provider, that difference matters at the shortlist stage. Ask for confirmation of the exact provider, region and database version you intend to use. A general statement that software can run anywhere leaves unanswered whether the supplier will operate and support your chosen configuration under the proposed agreement.
Keep the hosting decision distinct from your assessment of the database operator. We recommend documenting the entities providing infrastructure and database management, the locations from which support may work, and the information that leaves the database environment. Include backups, diagnostic logs and support attachments in that discussion, rather than considering only the primary database.
Consider a hypothetical manufacturer that wants its reporting database on European infrastructure but needs specialist help with maintenance. A supported deployment in its own account could meet that operational aim. The company should still check the management arrangement against its requirements for access and continuity before describing the whole service as sovereign.
Agree how the operator can access your systems
Management access is part of the service you are buying. The important questions are how broad it is, who can use it, and what happens when you restrict it. We would review infrastructure permissions, database access and emergency support procedures separately, because approval in one area should not stand in for an explanation of the others.
Altinity's access documentation makes a useful distinction between personnel access to database contents and permission to perform management operations. It describes controls for reading or changing data, alongside separate controls for configuration changes and actions such as upgrades or restarts. It also explains that customers may temporarily lift restrictions to enable support work.
Treat those controls as the beginning of the assessment. Ask the supplier to demonstrate the full access path, including cloud permissions and any routes used during an incident. Establish who approves temporary access, where the action is recorded and who checks that the previous restrictions have been restored.
Then walk through an urgent support case together. If only your team can approve a change, decide who is available outside normal working hours and what authority that person has. We would favour an explicit, rehearsed arrangement over access settings that look restrictive on paper but leave everyone unsure how to restore the service.
Give daily operations a named owner
The division of work changes with the deployment model. Altinity's responsibility guidance says customers running in their own environment take on some responsibilities the hosted service would otherwise handle. The split also depends on whether the customer supplies the cloud account alone or manages the underlying Kubernetes environment itself.
Before signing, ask for a written allocation of routine work and incident duties. It should cover database upgrades, infrastructure maintenance, monitoring, backups, security updates and capacity changes. For each activity, agree who performs it, who authorises disruptive changes and how your team verifies completion.
Use a familiar business event to test the answer. Suppose a scheduled upgrade coincides with the month-end reporting deadline. Your team needs to know whether it can move the maintenance window, how long deferral is supported, who checks application compatibility and what happens if the upgrade fails.
We recommend assigning one internal owner even when the supplier performs most tasks. That person coordinates approvals, checks that the support arrangement matches business hours and keeps the recovery procedure usable. For a lean IT team, the right arrangement is one whose retained responsibilities fit the people actually available to carry them out.
Prove that recovery and handover work
Recovery deserves its own demonstration. Altinity's backup setup guide for customer-managed Kubernetes environments requires configuring storage access and permissions. That is a concrete reminder to establish which backup setup tasks belong to you for the deployment you are considering.
Our suggested acceptance test is to restore a representative backup into a separate environment and run an application check against it. Record how long recovery takes and how recent the recovered information is. Confirm that the people expected to perform the recovery can obtain the required credentials, encryption keys and configuration without depending on one unavailable administrator.
Test supplier exit as a separate scenario. Altinity documents disconnecting a customer Kubernetes environment from its management service while leaving the ClickHouse database running. That is a useful capability, but its scope matters: continuing to run is only one part of taking over the service.
The company's more detailed handover guidance includes replacing supplier-managed connection addresses and traffic-routing components, arranging certificate automation and verifying administrative access. Parts of that procedure are specific to AWS, so request the corresponding plan for your chosen environment. The documentation makes clear why leaving deserves preparation even when the database files can stay where they are.
Ask your proposed operator to demonstrate a handover using a test system. Your team or a replacement operator should be able to connect the application, receive alerts, restore a backup and explain how maintenance will continue. Preserve the working configuration, software versions and instructions in a location your company controls.
Also distinguish changing operators from changing infrastructure providers. Keeping a database running in the same account does not demonstrate that it will work on another cloud. If a later move is part of your plan, include a separate trial on the intended destination and record any features or supporting services that need replacing.
Start with the control you actually need
We would assess BYOC against a specific objective. A business may need its own infrastructure account, a supported European hosting option or a practical way to change database operators. Each objective calls for different evidence, and none requires assuming that every managed database should move immediately.
Choose one representative application for a limited trial. Include ordinary queries and data imports, but also a maintenance request, an access approval, a restore and a planned handover. Agree the success criteria with the people who depend on the application, so that a technically successful trial also answers the business's questions.
If the service still depends on a specialist team for daily operation, make that dependency explicit and decide whether it is acceptable. If the retained work exceeds your team's capacity, reconsider the responsibility split or compare another hosting arrangement. The aim is an operating model your company can sustain.
Running a managed database in your own account can put infrastructure decisions closer to your team while preserving specialist help. The strongest evidence is a demonstrated ability to exercise the controls you need: choosing supported infrastructure, approving access, recovering the application and arranging an orderly handover. Those results give a board a clearer answer than account ownership alone.
Thinking about migration? Book a free consultation to discuss your situation.