TABLE OF CONTENTS
NVIDIA H100 SXM On-Demand
We’re back with your monthly dose of Hyperstack updates!
October’s shaping up to be a power-packed month at Hyperstack. With new OS images, enhanced account protection and upcoming Kubernetes features, your cloud just got smarter, safer and smoother. And yes, the best upgrades are still on the way.
New on Hyperstack Platform
Check out everything new on the Hyperstack this month:
New OS Images
You can now launch new VMs using AlmaLinux and Debian OS images on Hyperstack. These additions expand your choice of operating systems, giving you greater flexibility to match your preferred environments and workloads.
Account Lockout Protection
User accounts are now temporarily locked after multiple failed login attempts, adding an extra layer of protection against unauthorised access and ensuring your account remains secure.
Batch Node Deletion API
Managing clusters just got easier. The new Delete Multiple Nodes API lets you delete multiple nodes from a cluster in a single request, streamlining your operations and saving valuable time during scaling or cleanup.
Node Group Scaling Limits
With the introduction of the Update Node Group API, you can now define minimum and maximum limits for the number of worker nodes that the autoscaler (coming soon) can scale to within a Kubernetes node group.
Fixes and Improvements
We have made several improvements this month:
Improved Kubernetes Cluster Resilience
Control plane nodes are now deployed on separate underlying hosts (where available), reducing the impact of hardware failures and improving overall cluster stability.
Control Plane Stability for Autoscaler Compatibility
Control plane nodes are now labelled cluster-autoscaler.kubernetes.io/scale-down-disabled: "true" to prevent them from being removed by the autoscaler, ensuring essential infrastructure stays intact.
Cluster Reconciliation Behaviour
Cluster reconciliation now preserves any custom containerd configurations, so your manual runtime settings remain unchanged during updates or system reconciliation.
Updated Device Path for Volume Attachments
API responses now return persistent device paths (e.g., /dev/disk/by-id/virtio-...) instead of temporary ones (e.g., /dev/vda), improving reliability and scripting consistency.
Coming Soon on Hyperstack
We’re continuously working to make Hyperstack smarter and faster. Here’s a glimpse of what’s next:
-
Kubernetes Auto-scaling is coming soon, allowing you to automatically adjust resources based on demand.
-
AI Images: Gen AI Images are on the way to help you deploy and experiment with AI faster than ever. The images will allow you to deploy a specific LLM together with OpenWebUI.
- LoRA Adapter Imports: The ability to import LoRA adapters directly from Hugging Face is coming soon on Hyperstack AI Studio. Stay tuned to easily integrate fine-tuned adapters for custom and more flexible model inference.
Stay tuned for these updates in our upcoming rundowns.
New on our Blog
Check out our latest blog on Hyperstack:
Running AI Workloads:
Why Renting Cloud GPUs Beats On-Premise Hardware
Training and deploying AI models used to require heavy investments in on-premises hardware, racks of GPUs, storage, networking and dedicated teams. Now, cloud GPUs deliver the same power at scale, without the high upfront costs or maintenance. This shift makes advanced compute accessible for individuals, startups, and enterprises alike. In this blog, we’ll explore why cloud GPUs are the smarter choice for running AI workloads compared to traditional on-premise infrastructure.
Learn more in our latest blog.
Building Gen AI Doesn’t Have to Be Expensive:
How Hyperstack AI Studio Changes the Game
Everyone wants to build with Generative AI, from startups training niche chatbots to enterprises automating customer interactions. But for most, the excitement quickly fades when the costs of experimentation start to pile up. If you want to get started with Generative AI, you need to have more than a good dataset or a clever idea. It is about the infrastructure behind it including GPUs, model hosting, data pipelines and deployment setups. And that’s where things get complicated, really fast.
Learn more in our latest tutorial.
Have an idea you'd like to see in Hyperstack? Let’s bring it to life.
At Hyperstack, we’re committed to continuous improvement and your ideas are a key driver of our innovation.
→ Is there a feature you’ve been waiting for?
→ Something that could speed up your workday?
→ Or a tweak that would make things feel effortless?
Tell us what would make your Hyperstack experience even better. Your feedback sets the direction for what we build next.
Missed the Previous Updates?
Catch up on everything you need to know from Hyperstack Monthly below:
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
