Logo
FAQs

FAQs

Frequently asked questions within the community

Community Questions

Platform & Technology

Orion is a containerized workstation platform that allows you to deploy on-premises, in the cloud, or in a hybrid environment. Our platform provides density on rack through technologies like GPU time-slicing, along with cost savings via auto-scaling and aggressive scheduling. Juno is built on Kubernetes, making it cloud-agnostic and extremely flexible.

Orion is powered by three core features:
  • Kuiper – Orion’s optimization engine. It dynamically slices and schedules compute resources, including GPU, CPU, and RAM, to maximize utilization (up to 92%), increase user density, and reduce costs.
  • Terra – Orion’s deployment engine. It simplifies Kubernetes cluster setup, application deployment, scaling, and lifecycle management, bringing everything under one control layer for IT and DevOps teams.
  • Helios – Orion’s interface. It provides the user-facing workspace where teams interact with Orion. Helios delivers secure, customizable containerized workstations and desktops that launch in seconds, giving creatives, developers, and researchers fast access to GPU/CPU-powered environments.

Together, these features make Orion cloud-agnostic, production-proven, and designed to turn your infrastructure from a fixed expense into a competitive advantage.

Orion uses dynamic provisioning instead of static allocation. Traditional VDI pre-provisions fixed VMs that often sit idle, while Juno spins up containers on demand, packs them densely onto nodes, scales nodes up/down automatically, and tears them down when not in use—so you only pay for active compute.

Our preferred protocol is Kasm, which delivers exceptional performance including color accuracy for creative work. We also support RustDesk and NoMachine out of the box. Other protocols like PCOIP can be integrated, though we don't officially support them yet.

Orion typically reduces cloud costs by 60-90% through:
  • Elastic scaling: Spinning up/down compute in real-time—eliminating idle servers idle servers
  • GPU time-slicing: Co-locating multiple workloads on single GPUs
  • Node packing: Scheduling containers densely
  • Scale-to-zero: Powering off nodes when idle
  • Creative iteration: Faster workstation provisioning means less downtime between iterations, allowing teams to explore more creative options while maintaining tight deadlines
Deployment & Capabilities

Orion runs on any Kubernetes cluster—EKS, AKS, GKE or on-premises hardware. We're 100% cloud-agnostic and extraordinarily friendly for on-prem deployments, as everything is containerized and ready to go.

Windows support will be available (TBD) but with limitations due to Microsoft's terms of service. Unlike our Linux containers, Windows workloads must use a 1:1 relationship with the underlying hardware—we cannot offer GPU time-slicing for Windows. Windows licensing is strictly BYOL (Bring Your Own License), as we do not redistribute Windows licenses. We are actively working to improve Windows integration, though we don't currently have a definitive timeline for feature parity with our Linux offerings.

If container images are cached on your nodes, a new workstation is ready in 2-5 seconds. For uncached workloads requiring new nodes, provisioning takes 1-2 minutes for CPU workloads and 2-3 minutes for GPU workloads, depending on the cloud provider.

Absolutely. Orion simply mounts your existing file shares and toolchain locations into containers—you keep your existing pipelines no matter how complex or customized. This makes Orion ideal for studios with significant investment in proprietary tools.
GPU & Hardware Resources

No license is required for Juno's default GPU Time-Slicing mode—you simply run the standard NVIDIA GPU operator and share your GPU across multiple containers. If you want true hardware isolation (MIG), you'll need NVIDIA GRID/VGPU licensing—Orion fully supports both modes.

Yes—since time-sliced containers share the same physical GPU memory, oversubscribing VRAM (for example, if users collectively request more than the available GPU memory) can crash the GPU. Orion provides the flexibility to use time-slicing where appropriate and dedicated allocation where necessary. We strongly recommend thoroughly reviewing our documentation before configuring your GPU setup, as Juno is not responsible for hardware damage resulting from improper configuration or oversubscription of resources.

Orion mounts any CSI-compatible network storage (NFS, EFS, Qumulo, etc.) directly into containers. We've had excellent results with Qumulo's Cloud Data Fabric , which can be set up in approximately 15 minutes.

Yes! Orion can run on everything from Raspberry Pis to full enterprise-grade hardware. As long as it can boot and run containers, Orion will support it. This means you can repurpose older workstations and towers that might otherwise become e-waste by adding them to your Kubernetes cluster. While each workstation still runs as an individual container, pooling older hardware into a cluster helps maximize utilization of existing investments and extends the useful life of your equipment.
Security & Administration

Orion delegates authentication to external providers like Google OAuth, AWS Cognito, and soon Okta (which will bring support for Active Directory and LDAP). This approach ensures Juno never handles passwords directly.

Orion implements Role-Based Access Control (RBAC), allowing you to assign users to roles (Admin, Supervisor, Employee, etc), projects, and services. You can also configure fine-grained network policies per namespace or project. This flexible permission system works for organizations of all types, not just creative industries.

Yes—our Genesis dashboard lets administrators manage projects, invite users, assign roles, configure scaling policies, and install application plugins from our built-in Terra app store. Users can request and access their workstations through an intuitive interface.

Terra uses Helm charts and simple shell scripts. You can package any Linux application and publish it to our app store, giving users one-click installation. The framework is open-source, allowing for community contributions. Additionally, you can add your own private repository to your Terra store for internally developed applications that you don't want to share publicly.
Licensing & Support

Our licensing is simple: per user, per month, with volume discounts available for larger teams.

Not really. Most customers run Orion with a small operations team (1-2 people) thanks to our declarative API, Helm-based installations, autoscaling, and the Genesis dashboard for self-service management.

Yes, a proof-of-concept can be up and running in under 30 minutes. However, it's essential to first review the prerequisites in our  Orion Documentation  before beginning. Our setup process involves provisioning a Kubernetes cluster, installing Genesis via Helm, mounting your storage, importing users, and launching your first containers.

We recommend having someone on your team with basic Kubernetes knowledge before attempting installation. Currently, we provide installation via Helm charts (Ansible playbooks are in development but not yet available). While we've successfully tested deployment with non-technical users who managed with only minor issues, please note that comprehensive support is only available through our paid support packages. Our documentation is designed to guide you through the process step by step.
DisclaimerCookie Policy
Juno LogoClick to view the Juno Innovations linkedin© 2025 Juno Innovations All Rights Reserved
Privacy PolicyTerms & Conditions