Service description

As of: 14 October 2026

This English version is provided for convenience. Only the German version is legally binding.

This service description sets out what gh0stservice GmbH commits to for its customers. It has two chapters:

• Chapter Self-Service applies to every gh0stcloud account, for consumers and businesses alike.
• Chapter Managed applies in addition to Managed Apps and to applications we operate for you under an order form (Bestellschein). This chapter applies only to businesses within the meaning of § 14 BGB (German Civil Code). The exception is clause B.11 on platform extensions: it applies to every account, consumers included.

Clause 1.2 of the GTC sets the order of precedence: order form, then this service description, then the GTC, then the price list. The Data Processing Agreement prevails on the processing of personal data. Amounts in the Managed chapter are net, plus statutory VAT. The amounts in the current price list apply. The Managed chapter sets out the service level terms in revision 2026-10-14.1. The portal names this revision with every order.

Chapter Self-Service

A.1 What we operate and what you operate

1. We operate the platform: data centres, clusters, servers, Kubernetes and its updates, delivery from your Git repository, public reachability and certificates, network rules, secret management, platform monitoring, the portal, the API and gh0stcli.
2. You operate your applications. That covers the code, images, charts and configuration in your repository, the databases and services you create yourself, their versions and updates, and the monitoring of your applications.
3. Projects under this chapter are labelled Operated by you in the portal.

A.2 Support

We usually answer support requests within 24 hours on business days (best effort). Business days are Monday to Friday, excluding public holidays in North Rhine-Westphalia. This means the first answer from a person, not the resolution. It applies to every severity.

A.3 Availability

We run the platform on a best-effort basis and do not commit to a specific availability. We publish the measured availability of the portal, the API, sign-in and cluster reachability on our status page.

A.4 Data backup

1. You operate your applications yourself. We do not offer data backup as a service for your data. You set up and are responsible for backups, restores and restore tests, for example with the backup feature of your database operator and a storage target of your own.
2. The platform has no backup feature for this data. Mirrored storage protects against the loss of a single machine, but it is not a backup.
3. Copies we keep so that we can recover the platform itself are not a service to you. We promise nothing about them and do not restore them on request.
4. If you want us to back up your data, book a Managed App or have us custom operate the application under the Managed chapter.

A.5 Maintenance

1. Disruptive platform work happens only in the staffed maintenance window, Tuesdays 06:00 – 09:00 (Europe/Berlin). We announce it 14 days ahead on the status page.
2. Outside business hours and the staffed maintenance window, only automation acts. We may still carry out urgent security work voluntarily. It carries no response commitment, and we inform you afterwards.

A.6 Location

The platform runs in data centres of Hetzner Online GmbH in Germany. Each cluster sits in one location and is reached through one load balancer. We do not commit to recovery in another location.

A.7 Help by the hour

Businesses can book advice and hands-on help by the hour, for example for a migration, troubleshooting or a restore of their own. You remain the operator of your application. We do not provide ongoing operation by the hour; ongoing operation is only available under the Managed chapter.

A.8 Leaving

We charge nothing for switching provider. You export your data with your own tools, and we do not charge for the data transfer this needs while you switch. The details are on the page Switching providers and data portability.

Chapter Managed

B.1 What we offer

1. Managed App. An application from our catalogue which we run for you, for example a database such as PostgreSQL or MariaDB, or a web application such as the password manager (Vaultwarden). Vaultwarden is an independent project and not affiliated with Bitwarden, Inc. You order it in the portal or through gh0stcli. We offer a package, starting size or service level only once we have released it. These projects are labelled Managed App in the portal.
2. Custom operated by gh0stservice. Any other application we operate under an order form (clause B.10). These projects are labelled Custom operated by gh0stservice in the portal.
3. Both come with two service levels: Managed Basic and Managed Business.
4. Size and cost. Managed Apps have no fixed plans. You choose the size of your app (processing power, memory and storage) freely and can change it at any time; storage can be increased but not decreased. Resources are billed by the reserved amount, measured to the minute, at the rates of the price list in force. What is fixed is the service package of SLA and support that you acquire with the chosen service level; its service fee follows the price list and may be staged by the size class of your app. If you reduce the sizes, the size class of the service fee stays until the end of the minimum term. The SLA applies to the sizes you have chosen; load beyond the chosen sizes is excluded from the commitments.
5. This chapter applies only to businesses with a payment method on file; clause B.11 applies to every account. Anything this chapter does not cover is governed by the Self-Service chapter.

B.2 Definitions

• Business hours: Monday to Friday, 09:00 – 17:00 (Europe/Berlin), excluding public holidays in North Rhine-Westphalia. There is no round-the-clock on-call service.
• Business hour: one hour within business hours. Business day: a day with business hours.
• Staffed maintenance window: Tuesdays 06:00 – 09:00 (Europe/Berlin), with one person on duty.
• First qualified response: a solution, an analysis or a concrete next step from a person. An automatic acknowledgement does not count, and the response is not the resolution.
• Order confirmation: our acceptance of your order. It forms the contract for the instance and states the service level, the starting size, the fee rule and that the service fee starts on the handover day.
• Handover notice and handover day: The handover notice supplements the order confirmation and states the handover day. We send it in text form, and only once the instance has shown successful backups that we observed and we have run and recorded one restore.
• Instance: a single operated application with its data, for example one database, or a web application with its database and its files.
• Size: the amount of processing power, memory and storage you choose for an instance. You can change it at any time; storage can be increased but not decreased.
• Starting size: an initial size offered in the portal (S, M or L) that an instance can begin with. It is not a fixed plan, and every size can be changed afterwards.
• Size class: the step of the price list by which the service fee may be staged. It follows from the sizes you choose.
• Service package: the commitments of a service level (SLA) and the support under clause B.3.
• Service fee: the monthly fee per instance for the service package of the chosen service level.
• Resource cost: the charge for the processing power, memory and storage of an instance. We bill it by the reserved amount, measured to the minute and invoiced monthly.
• Restore start: we have created the job that restores the requested restore point into a new instance. We report when it completes; we do not commit to a duration.
• Recovery point design target (in the order: design target for the recovery point): the maximum intended age of the newest recoverable point: 24 hours for Managed Basic, 6 hours for Managed Business. The target describes what our work is aimed at. It is not a guarantee of success.
• Newest recoverable point: for MariaDB, the newest successful full or incremental backup. For PostgreSQL on Managed Basic, the newest successful base backup. For PostgreSQL on Managed Business, the last archived segment of the transaction log (WAL) on top of a valid base backup that is at most 26 hours old. Where a Managed App stores files besides its database, for example attachments or documents, its newest recoverable point is the older of two points: the database point under the sentences above, and the newest successful backup of the files.
• Data inside the recovery point design target unrecoverable: every successful backup from within the design target window failed to restore, including one retry of the newest backup within 1 business day. Every attempt is recorded as a job.

Severity classes:

We assign each report to a class and may reclassify it with a stated reason. If you report an incident as critical, please describe the impact on your business. We treat a security or personal data breach as critical; the notification periods of the Data Processing Agreement apply in addition.

B.3 Commitments per service level

We measure response times from the timestamps of your report within business hours. The clock stops while we are waiting for your reply. Reports about the same platform incident declared on the status page share one first response.

The backups of an instance are kept in a storage area of their own assigned to your organisation alone. It has its own credentials and is kept apart from the storage areas of other customers.

B.4 Updates

1. We apply patch updates, meaning fixes within the same minor version, in the instance's maintenance window. We do not ask for consent to each of them.
2. Upgrades to a new minor or major version happen only with your consent.
3. We may apply urgent security updates outside the maintenance window if waiting would put security at risk. We inform you afterwards.

B.5 Restores

1. You choose the restore point within the retention period. We start the restore within the time of your service level.
2. A restore always creates a new instance. We do not overwrite the existing instance. The new instance is a new, separately billed instance.
3. The new instance keeps the credentials of the source, including the password your applications sign in with. Connected applications can therefore switch over to the new instance without new credentials.
4. You approve the switch to the new instance and check the data afterwards.

B.6 Contract, handover and fees

1. The contract for an instance is formed by our order confirmation.
2. The service fee starts on the handover day. In the first month we charge it pro rata by day.
3. The minimum term is 3 months from the handover day. After that, the contract for the instance can be terminated at any time with effect from the end of a month. If it ends earlier, you owe the service fees for the rest of the minimum term. We tell you this amount before you order.
4. We invoice the Managed Business setup fee and the setup fee for a custom operation on the handover day. If handover has not happened within 30 days of the order confirmation for reasons we are responsible for, you may withdraw from the order without charge.
5. A move to a higher service level applies at once. We charge its service fee from the first day of the month after the month in which we first observed the new service level's backup cadence on the instance. Until then the current service fee applies. When you move from Managed Basic to Managed Business, we charge the Managed Business setup fee once, unless it has already been paid, in the month the new service fee starts. The service fee of a higher size class applies from the following month. A move does not restart the minimum term. A move to a lower service level is possible at the end of a month once the minimum term has ended; the lower service level and its service fee apply from the first day of the following month. If you reduce the sizes, the size class of the service fee stays until the end of the minimum term and then follows the sizes you chose. We bill the resources by the new reserved amount from the moment of the change.
6. The price list may set a fee of its own for a package, per service level, per size class and for add-ons. Without such a package line, the service fee of the service level applies.
7. The base fee, the resource cost and the data transfer of the instance are charged in addition, per the price list. We bill processing power, memory and storage by the reserved amount, measured to the minute and invoiced monthly. Amounts we give you before an order or a change are estimates for the chosen sizes; what is invoiced is the amount actually reserved.

B.7 Credit

1. If, in a calendar month, the newest recoverable point was at any time older than twice the recovery point design target (Managed Basic 48 hours, Managed Business 12 hours), or we missed the response time for a critical incident, you receive a credit (a flat-rate price reduction) of 25 percent of that month's service fee for the instance.
2. If data inside the recovery point design target cannot be restored, the credit is 100 percent of that month's service fee.
3. There is at most one credit per instance and month; if both cases apply, the higher rate counts. The base fee, resource costs and hours are not affected.
4. We apply the credit to your next invoice without being asked. Any remainder is refunded when the contract ends.
5. The credit is offset against statutory price reduction (Minderung) and damages claims for the same event. Further statutory claims and rights, in particular to extraordinary termination, remain unaffected.
6. If three months within twelve are affected by a credit, you may terminate the contract for that instance for cause.

B.8 Exclusions and what we do not promise

1. The following do not count against the commitments or the credit: announced maintenance; changes outside the application's parameters that you requested, and incidents causally related to them; load beyond the chosen sizes, or a full disk after you ignored a warning; declined updates; upstream defects for which no fix is available; suspension for non-payment.
2. Not included: your application code, the design of your database schema, the tuning of your queries, and support for your end users.
3. Not promised: restores into the existing instance, a time until a restore completes, a recovery time objective (RTO), an availability target for single instances, and round-the-clock response.

B.9 Your part

• You name a person who can be reached during critical incidents.
• You choose the restore point, approve the switch and check your data after a restore.
• You are responsible for database users, grants and the correctness of your data.
• You support your own users with their questions.
• You consent to or decline planned upgrades within a reasonable time.

Delays caused by missing cooperation extend the periods in this chapter accordingly.

B.10 Custom operated by gh0stservice

1. An application without a catalogue package is operated under an order form. The same two service levels and the same commitments apply as for Managed Apps.
2. The order form states a fixed amount for setup.
3. The order form contains a table with the scope for this application: which components we operate, which data the backups cover, how we verify a restore, and what is out of scope.
4. Handover, the start of the service fee, the minimum term, the setup fee and the credit follow clauses B.6 and B.7.
5. If the order form grants you co-administration, the commitments do not apply to incidents caused by your changes.

B.11 Platform extensions

1. The platform extensions are gh0steye (monitoring with metrics, logs and traces) and gh0stforge (Git, CI and a registry). They are part of the platform, not Managed Apps.
2. They run under the platform terms of your account: we usually answer support requests within 24 hours on business days (best effort). Availability is best effort. There is no service level for platform extensions.
3. We look after an extension's versions, configuration and maintenance. We back up an extension's data daily and keep the backups for 14 days. We restore it on request. There is no choice of version and no configuration beyond the extension's settings.
4. We charge the extension's measured usage per the price list. There is no extension fee.
5. An extension is switched on only after an owner of your organisation has agreed to its cost.
6. You remain responsible for your own content in an extension, for example dashboards, repositories, pipelines and the code they run.

B.12 Stand-in and escalation

1. Falk Hähner is responsible for operations. If he cannot be reached, Linus Stiehm takes over.
2. If neither can be reached, monitoring and automatic recovery keep running. To meet the response times, we may bring in an experienced external operations professional. They get access to personal data only with your consent under clause 5.3 of the Data Processing Agreement. You can also give this consent in advance for incidents.

B.13 Additional services

Work outside the service level, for example changes outside the application's parameters, migrations or major upgrades on Managed Basic, is charged by the hour per the price list, currently EUR 120.00 net per hour in 15-minute increments. For migrations we give you a written estimate first. Work outside business hours only by prior agreement.

B.14 Export and end of contract

When you leave, you receive for each Managed App, at no extra cost, an export in the application's native format: a database dump and, for web applications, the files as well. Clause 12 of the GTC and the page Switching providers and data portability govern switching and data transfer.