TABLE OF CONTENTS
NVIDIA H100 SXM On-Demand
Hyperstack Monthly Update
It’s time for our February monthly update.
February at Hyperstack was focused on making the platform more resilient and predictable. We introduced better safeguards for Kubernetes environments, improved VM deployment reliability and enhanced networking defaults.
Scroll down to see what we shipped this month!
What’s New on Hyperstack
Here's what we released on Hyperstack in February:
Firewall Warnings for Kubernetes VMs
We’ve added a warning banner to the Firewalls tab for VMs in a Kubernetes cluster. This banner alerts users before making firewall configuration changes that could disrupt:
- Cluster networking
- Reconciliation processes
Image-Flavour Compatibility Support
Hyperstack now supports detecting compatibility between selected images and VM flavours. While no images currently define restrictions, this enables:
- Future warnings for suboptimal image
- Restrictions on incompatible configurations
Latest Fixes and Improvements
- Security Enhancements: Platform security has been further strengthened with internal improvements that increase protection and provide safer, more reliable system interactions for users and their workloads.
- Improved Usability and Error Messaging: User experience has been improved with clearer error messages and more intuitive system feedback, making it easier to understand issues and resolve them quickly.
- New “No-Reboot” Flavour Label: Applicable flavours now include a “no-reboot” label, helping users easily identify configurations that support operations without requiring instance reboots—reducing downtime and maintaining workload continuity.
- Reliable VM Creation with Bootable Volumes via API: VM creation using the Create Virtual Machines API with
create_bootable_volume = truenow waits for the volume to become available before proceeding, improving provisioning reliability. - Firewall Management Fixes: Resolved issues with firewall assignment and detachment workflows. The VM list in the firewall assignment modal now correctly shows only VMs in the same environment as the selected firewall, and the “Save changes” button now functions as expected when removing all firewall attachments from a VM.
- Improved Redirect After Session Expiry: Users are now correctly returned to the page they were on after signing in again, even if their access token was missing or had expired during refresh.
- Fixed TLS SAN Mismatch in Standard Kubernetes Deployments: The public IP address of the master node serving as both bastion and API endpoint (in Standard configuration deployments) is now included in the Kubernetes API server certificate’s Subject Alternative Names (SANs), resolving TLS handshake failures caused by hostname mismatches.
- New Default Networking Mode for Kubernetes Clusters: New Kubernetes clusters now use VXLAN encapsulation (UDP port 4789) for pod networking instead of IPIP, allowing firewall configuration without disrupting connectivity. Existing clusters remain on IPIP.
- Environment VM Limit Notification: Users are now notified during VM or cluster deployment if the selected environment has reached its maximum VM capacity, with guidance to deploy in a different environment.
New on the Blog
Check out exciting blogs on Hyperstack this month:
How to Deploy Ollama on Hyperstack:
A Quick Setup Guide
Our Ollama setup guide shows you how to deploy Ollama on Hyperstack so you can quickly run LLMs on GPU-powered cloud infrastructure. Ollama is ideal for fast experimentation and local-style model testing while Hyperstack provides on-demand GPUs for reliable performance. Follow the steps in the tutorial to launch a working Ollama setup in minutes and start running models with minimal configuration.
Learn more in our latest blog.
5 Hidden Costs of Picking the Wrong GPU Cloud Provider
What You Need to Know
Teams building AI, LLMs, GenAI, ML pipelines and HPC workloads face pressure to ship faster, control costs and still meet security, compliance and performance requirements. Many GPU cloud providers fall short and choosing the wrong one can quietly drain budget. The real cost rarely matches the pricing page but it shows up later as delays, high bills, weak security and frustrated teams. In this blog, we break down five hidden costs of the wrong GPU cloud platform and what modern GPU infrastructure must deliver to run AI at scale.
Learn more in our latest blog.
How to Choose the Right Generative AI Platform:
For Your Projects
Most Generative AI projects don’t fail because the model underperforms. They fail because the platform you deploy on cannot support the full Gen AI lifecycle. You may have a strong idea and a powerful model but they are not enough. What makes it a success is how fast that idea can move from prototype to production. If each step requires a different tool or custom infrastructure, speed drops. This is why you need a single platform that offers it all. In this blog, we help you understand the factors important to choosing the right generative AI platform for your projects.
Learn more in our latest blog.
Your Ideas Power Hyperstack
You know your workflow better than anyone. If there’s anything you wish Hyperstack did differently or better, now’s your chance to tell us.
Maybe it’s a feature you’ve been thinking about, a tool that could speed up your workflow, or a simple improvement that would make your project easier. Whatever it is, we’re listening.
For any questions or suggestions, feel free to reach out at support@hyperstack.cloud. Stay tuned for even more updates and exciting tools next month.
Subscribe to Hyperstack!
Enter your email to get updates to your inbox every week
