Service Level Agreement
gridscale SLAs.
Definitions
- Reaction time describes the time window between when a qualified disruption has been reported via the customer, either by telephone or in writing to gridscale.
- POC Products that have been made available, before general avaialability to a selected customer group. These are defined by the agreement to test each product individually.
- Labs Products that are available generally, to all customer but still within a research phase. These are marked by “Labs” labels within the panels / API documentation.
- Deprecated Products As we offer managed services and follow the upstream of such provided services, we expect customer to update within due time. If not these versions will be marked as deprecated/outdated via the panels or API documentation.
Validity
The SLAs described in this document do not apply to services disrupted by:
- Legal or regulatory requirements, including court orders of the respective country in which the operation takes place.
- Overloading or false use of the services provided by gridscale for the customer.
- Changes to a service or configuration made by the customer.
- Maintenance work done by gridscale.
- Blocking of individual services or an entire account due to misuse or improper use of the account.
Disruption levels
| Disruption level | Description |
|---|---|
| Critical | Your service is no longer available |
| Significant | Your service is experiencing periodic disruptions or performance issues |
| Low impact | Your service appears to be working fine, with little to no customer facing impact |
Service Level Agreements
The service level agreement describes the minimum amount of uptime that we offer per product.
gridscale’s Services
gridscale offers the following types of services and SLAs
| Service | SLA | Description |
|---|---|---|
| Primary | 100% | CPU, RAM, Network Cards (NIC), Distributed Storage, Upstream, SDN (Private Networks) |
| Controlling | 99,999% | All APIs, Panels and tools used to access the platform |
| Platform Services | 99,99% | gridFs, gridStore, gridSQL, Load Balancer, Managed Kubernetes (GSK), Object Storage |
| Secondary | 99,9% | All other services not mentioned above |
| Deprecated | 0% | All Deprecated products e.g Security Zones, outdated Platform Services, POC, Labs |
Discontinued
Discontinued means that a product is no longer being developed or optimized by us.
This can affect any product in the Primary, Controlling, Platform Services or Secondary categories. The SLA for that category will remain in place for the discontinued product.
Reaction Times
gridscale guarantees reaction times, base on the disruption level
| Disruption level | Reaction Time |
|---|---|
| Critical | 45 minutes |
| Significant | 180 minutes |
| Low impact | Best Effort |
SLA breach
If the SLA is breached, a credit note is issued automatically. In the case of a time-based measurement, the credit note is made up of the duration of unavailability, rounded up to 5 minutes, multiplied by the fees incurred for all directly affected services.
Example
The credit for the outage of a virtual server is calculated as follows:
- Unavailability of 16 minutes (rounded up to 20 minutes)
- Price of the server: 1.20 EUR/minute
- 20 minutes x 1.20 EUR = 24.00 EUR credit note
Maintenance
gridscale runs a zero-downtime architecture which means that an overwhelming majority of updates do not affect services. If an update will affect a component, and more than one customer at a time it will be announced 3 business days in advance by gridscale on the status page.
Maintenance on individual components like dedicated hardware will be agreed on with the customer before hand.
Maintenance or force-upgrades on Platform Services are announced via email and within the notification center of the Partner and Cloud Panels at least 3 weeks before the maintenance should happen, and if possible a time window is mutually agreed upon between gridscale and the customer.
Events can occur which require a no notice maintenance like critical security patches for example, these are handled as primary incidents, if they cause any disruption to services.