VM (Virtual Machine)
Linux / Windows VMs for databases, middleware, legacy business and test environments.
- Snapshot / backup / restore
- PVE cluster scheduling
- Auto-reclaim on expiry
Product · Shuke Smart Cloud
Put AI-OS in your datacenter,
compliance and compute solved together.
Shuke Smart Cloud pools VMs, containers, object storage and AI/GPU into one base. PVE + Kubernetes dual foundation. OpenBao secrets, department isolation, quota approval, ops audit — AI-OS and AiDo run in your house as smoothly as in public cloud.
Resources · 4 types
One pool. Compute, storage, network, AI — all included.
Pool existing servers into VM / containers / storage / AI. Self-service via Portal, unified approval and reclaim.
Containers
Kubernetes namespaces for APIs, microservices, scheduled tasks, internal platforms.
- Namespace quota isolation
- Helm / CRD / GitOps
- One-click observability
Storage
Block, object, shared file. Serves databases, backups, archives.
- S3-compatible object storage
- Multi-replica block storage
- Cross-node shared FS
AI / GPU
Inference instances, vector services, embedding, training — unified scheduling entrance.
- GPU quota by department
- Mido inference gateway
- Train / infer hybrid scheduling
Foundation · Dual base
PVE for virtualization, K8s for cloud-native. No collision.
Care for legacy VM business as well as new cloud-native apps — both bases sit under the same Portal control plane.
PVE virtualization base
Proxmox VE multi-node cluster. VMs and persistent storage. All lifecycle managed by Portal.
- VM create / delete / power
- Spec adjust / snapshot
- Backup / restore
- Renewal & auto-reclaim
Kubernetes base
K8s multi-cluster, namespace-based quota distribution. Cloud-native workloads and AI inference.
- Namespace creation
- Quota management
- Mido inference gateway
- GitOps & Helm
Security & Observability
Compliance baseline for government and enterprise.
Secrets, permissions, approvals, audit, observability — five things made into ready-to-use configuration.
OpenBao secrets
VM passwords, API keys generated and destroyed automatically by OpenBao.
Department / project isolation
RBAC + quotas, 3 levels of access: department, project, individual.
Approval & audit
High-risk ops (delete, restore) go through human approval + full audit log.
Unified observability
Prometheus / Loki / Jaeger collect together; alerts and traces reach the business layer.
Deployment · Landing path
Four steps, from datacenter to AI-OS.
Global DT delivers solution and implementation. From hardware inventory to AI-OS onboarding to handover — we don't make customers do both sides of the contract.
Requirements survey
Audit existing datacenter / network / storage / compute. Define compliance and AI-OS landing scope.
~1-2 weeks
Foundation build-out
Deploy PVE + K8s dual base, integrate storage, network, observability, OpenBao, audit log.
~2-4 weeks
AI-OS / AiDo onboard
Deploy AI-OS and AiDo Studio onto private cloud. Set up departments, quotas, Agent templates and release approvals.
~1-2 weeks
Handover & ops
Ops SOP, on-call, DR drills handed to customer team. Global DT keeps continuous upgrades and yearly health checks.
Ongoing
FAQ
Budget / compliance / hardware / SLA — five common questions.
Does it support PRC Cybersecurity Classified Protection 2.0 Level 3?
Yes. The private cloud is designed to PRC MLPS 2.0 Level 3. Key vault, audit logs, separation of duties, backup-restore all conform. Customer-specific hardening is part of every project.
Can we keep existing hardware?
Yes. PVE + K8s dual base supports mainstream x86 servers and domestic platforms (Phytium, Kunpeng). Common RAID cards, NVMe SSDs, 10G/25G NICs are validated. GPU includes NVIDIA A/H series and domestic inference cards.
Must AI-OS run on private cloud?
No. AI-OS provides public, dedicated and private cloud deployments. Customers can start with public cloud pilots and move core data / models to private later — hybrid is fine.
SLA? Incident response speed?
Standard SLA: control plane 99.9%, single VM 99.5%. P1 incidents responded within 15 minutes, restored within 4 hours. Enhanced SLA with on-site or dedicated on-call is available.
Can we just buy 'the resource pool' without AI-OS?
Yes. Private cloud can run standalone as general infrastructure; you can onboard AI-OS / AiDo on your own schedule. We recommend 'base first, platform next, agents last' to avoid heavy upfront cost.
Get in touch
Plan a session first, decide together later.
Private cloud is a serious investment. Global DT always offers a free 1-2 hour planning session first — we walk through your datacenter, business and compliance state before recommending anything.