Cloud Architecture Insights

Uptime is defined as the percentage of time the core platform APIs and orchestration layer are accessible and operational, measured over a calendar month. Scheduled maintenance windows, communicated at least 72 hours in advance, are excluded from this calculation. Our 99.95% SLA applies to the control plane, not individual customer workloads or data plane performance.

Data Residency refers to the physical geographic location where customer data is stored at rest. Data Sovereignty refers to the legal jurisdiction and laws governing that data. We offer region-specific clusters (e.g., US-East, EU-West) to meet residency requirements, but ultimate sovereignty is determined by the laws of the country where the data center is located. Clients are responsible for configuring their workloads accordingly.

Our "Managed" service includes automated provisioning, patching, security updates, and 24/7 monitoring of the Kubernetes control plane (master nodes). The "Self-Service" tier provides the raw cluster; the client is responsible for node maintenance, security patching, and control plane operations. Application deployment, scaling, and management within the cluster are always the client's responsibility unless covered under a separate professional services agreement.

Our platform provides infrastructure resilience through multi-zone availability and automated failover for hosted services. However, comprehensive Disaster Recovery (DR) encompassing application state and data recovery is a shared responsibility. We manage the recovery of the platform itself. Client data backups, application-level replication, and DR runbook execution are the client's duty. We offer tools and blueprints but not automated execution unless contracted.

Burst Scaling allows automatic, short-term allocation of additional compute resources beyond the committed baseline to handle traffic spikes. This is triggered by predefined metrics (CPU, memory, queue depth). Costs for burst resources are billed per-second at a premium rate, detailed in the resource appendix. Burst capacity is not guaranteed and is subject to overall cluster capacity at the time of the event. Sustained usage will trigger recommendations for baseline commitment increases.