SLAs¶
Per-service availability targets and credit terms.
Availability targets¶
| Service | Target (monthly) | Notes |
|---|---|---|
| Virtual Machines | 99.95% | Multi-AZ deployment required |
| Bare Metal | 99.9% | |
| Block Storage (NVMe HCI) | 99.99% | |
| Block Storage (Provisioned IOPS) | 99.99% | |
| Object Storage (S3) | 99.95% | Per region |
| Object Storage (Archive) | 99.9% | |
| File Storage | 99.95% | |
| VPC, LB | 99.99% | |
| DNS | 100% | Anycast across 3 sites |
| Managed PostgreSQL HA | 99.95% | HA topology |
| Managed Redis HA | 99.95% | |
| Managed Kubernetes — control plane | 99.95% | HA |
| Inference Endpoints — shared | 99.5% | |
| Inference Endpoints — dedicated | 99.9% | |
| Backup-as-a-Service | 99.9% (job success) | Successful backup runs |
| DRaaS — replication lag | RPO target per tier | |
| DDoS Premium — time-to-mitigate | < 60s / 5 min / 15 min | See service page |
Credit terms¶
When a service misses its SLA in a given month, you receive a service credit as a percentage of that service's charges for that month:
| Monthly availability | Credit |
|---|---|
| < target but ≥ 99.0% | 10% |
| < 99.0% but ≥ 95.0% | 25% |
| < 95.0% | 50% |
Credits apply to the affected service, in BDT, on your next invoice. Maximum credit per month is 50% of the affected service's monthly charge.
Exclusions¶
- Customer-induced outages (misconfiguration, exhausted quota, IAM denials)
- Force-majeure events as defined in the MSA
- Scheduled maintenance announced ≥ 7 days in advance
- Single-AZ workloads where multi-AZ is required for the SLA to apply
How to claim¶
- Open a ticket within 30 days of the affected month
- Reference the affected resource IDs and the time window
- Cloud Digit reviews against telemetry; credit issued on next invoice