Service Levels
This page describes GTVA’s general approach to service levels for the operation and support of client infrastructure. It is for information only and does not constitute a service level agreement. Specific, binding service levels are agreed individually in each client contract.
Our approach
For clients whose clusters GTVA operates or supports, we structure service levels around what actually matters for AI infrastructure: keeping the system available and catching problems before they cost compute time. Depending on the engagement, that typically covers:
- monitoring of infrastructure health, around the clock where agreed;
- defined incident acknowledgement and response targets, graded by severity;
- response, mitigation and restoration targets, with resolution defined in the applicable contract;
- coordination of hardware replacement and warranty claims;
- maintenance of firmware, configuration and health-check tooling;
- where included in the engagement, monitoring for significant performance degradation and abnormal behaviour, not only complete outages;
- regular reporting on availability and incidents.
Agreed individually
The specific figures — acknowledgement, response and restoration targets, availability commitments, coverage hours, spare-handling arrangements, and what is measured — depend on the size of the cluster, its criticality and the client’s own requirements. They are set out and made binding in the service contract for each engagement, not on this page.
Contact
To discuss service levels for a specific deployment, get in touch through the contact form.