"Nebula" in the context of Kubernetes usually refers to one of three different technologies: OpenNebula (infrastructure platform), Nebuly nos (AI GPU optimization), or Nebula Graph (database). Based on common Kubernetes (K8s) use cases, here are the main options:
1. OpenNebula (Kubernetes Engine - OneKE)
OpenNebula is an open-source enterprise cloud platform that provides a certified Kubernetes distribution designed for edge and on-premise environments.
- OneKE (OpenNebula Kubernetes Engine): A CNCF-certified Kubernetes distribution, often using SUSE Rancher's RKE2.
- Key Features: Automated lifecycle management, integrated networking (SDN), and storage (Longhorn/LVM).
- Best For: Running hybrid clouds, edge computing, and managing virtualized and containerized workloads on a single platform.
- Architecture: Validated with Cluster API, Cloud Provider Interface (CPI), and CSI.
This video explains how to deploy a Kubernetes cluster using OpenNebula:
2. Nebuly nos (AI Optimization)
Nebuly nos is an open-source "operating system" module designed to optimize GPU utilization on Kubernetes.
- Key Features: Dynamic GPU partitioning (shares one GPU across multiple pods), elastic resource quota management, and performance-based scheduling.
- Best For: AI/ML workloads on K8s to reduce costs and increase GPU efficiency.
3. Nebula Graph Operator (Database)
Nebula Graph is a distributed graph database that runs on Kubernetes.
- Key Features: Automated deployment, scaling, and self-healing of Nebula Graph clusters (metad, storaged, graphd) using the Nebula Operator.
- Best For: Deploying high-performance graph databases on top of existing Kubernetes infrastructure.
Summary Table
| Technology | Focus | Use Case |
| OpenNebula (OneKE) | Infrastructure/Distro | Building Private/Edge K8s Clouds |
| Nebuly nos | GPU Utilization | AI/ML Workload Scheduling |
| Nebula Graph | Database/Storage | Running Graph DBs on K8s |
No comments:
Post a Comment