At Rackspace Spot, we're committed to delivering the ultimate managed Kubernetes experience at unbeatable prices. While our auction-based compute model has revolutionized cluster costs, we've known for some time now that we needed to improve the storage experience in Spot.
Today, we're excited to share beta availability of Rook/Ceph as our next-generation Container Storage Interface (CSI) provider in Spot. A new StorageClass, ‘spot-ceph’ is now available in Spot’s IAD and ORD data-centers. Specifying this storage class when using PVCs will use the ceph-csi driver to mount your persistent volumes on managed Ceph clusters in Spot.
Listening to Our Community
Our community has helped us understand the real-world experience with persistent storage in Spot. Through discussions on our GitHub repository and feedback from our users, we've identified several pain points with our current Cinder CSI implementation:
- Provisioning delays: Multiple users have reported issues with volumes stuck in provisioning states, with discussions like "Pods stuck in pending due to PVC" and "Persistent Volume stuck in "attaching" status"
- Inconsistent feature availability: The Cinder CSI experience varies depending on the underlying storage backend, making it difficult to provide a consistent experience across all our regions over time
- Operational experience: Users have often struggled to diagnose and self-help when there are situations such as stuck volume mounts, whereas a more “Kubernetes native” operating model would have made this easier and more intuitive
Why Rook/Ceph Matters for Developers
Our decision to invest in Rook/Ceph wasn't just about fixing problems. We also wanted to improve the storage experience for every user on our platform:
Faster, More Reliable Provisioning
Spot will continue to provide a dedicated storage service, using the existing storage infrastructure in the undercloud (based on Rackspace OpenStack). However, raw undercloud storage volumes are now aggregated via dedicated Ceph clusters (managed by Rook), which you can consume in your Spot cloudspaces via the ‘spot-ceph’ storage class. This model presents several advantages:
- The simplicity of a fully managed storage service: Spot is still providing fully managed storage infrastructure, you’re not responsible for managing storage uptime and availability
- Intuitive experience: Clear, Kubernetes-native consumption model, and Kubernetes-native error messages when something goes wrong
- Improved performance and reliability: Because the Rook/Ceph CSI is widely used in the cloud-native ecosystem, we expect volume mounts to be much more performant and reliable
Features That Just Work
Rook/Ceph provides multiple advantages over the older Cinder/CSI storage option:
Kubernetes-Native Operations
Rook brings storage management into the Kubernetes ecosystem, enabling:
- CRDs for everything: Manage storage pools, replication, and policies using familiar Kubernetes resources
- Better observability: Storage metrics and events flow through standard Kubernetes monitoring
- Simplified troubleshooting: One stack to understand, not two (Kubernetes + OpenStack)
Real-World Impact
Rook/Ceph is one of the most widely adopted storage solutions in the Kubernetes ecosystem, battle-tested across thousands of production deployments. This means:
- Better documentation: More examples, guides, and community resources
- Active development: Regular feature updates and CSI spec compliance
- Proven reliability: Used by organizations running mission-critical workloads globally
Let's talk about what this means for common workloads:
Running Production Workloads
With more reliable storage volume access, running production workloads becomes more realistic on Spot. This is especially important for Spot given the possibility of worker node pre-emption due to market price changes.
Stateful Applications
ReadWriteMany support means you can run applications that need shared storage across multiple pods—something that's not been available with our current Cinder CSI.
CI/CD Pipelines
Faster provisioning and more predictable performance mean your builds don't get stuck waiting for storage, and you can rely on snapshots for quick rollback strategies.
Addressing the Elephant in the Room: Reliability
We know what you're thinking: "What about reliability?" Frankly, this was one of the reasons we didn’t jump into Rook/Ceph before first tuning our existing storage performance to the best extent possible. Having done that, we feel that architecturally, Rook/Ceph gives us a path towards higher reliability, and that’s why we’ve invested in making this feature available.
Storage is just one piece of cluster reliability. We're making other investments in Spot to improve performance and reliability:
- Improved control plane performance and resilience: Spot Control plane infrastructure is being moved to a new architecture, v3, that will provide improved API server performance and reliability
- Enhanced monitoring: Better detection and alerting for storage and control plane issues before they impact workloads
- Faster problem response: Kubernetes native operations enable more self-service, and we’ll be monitoring storage health 24/7
The Rook/Ceph offering is part of a broader commitment to Spot platform reliability, not a one-off.
What This Means for Pricing
Here's the best part: our commitment to being the most cost-effective Kubernetes platform doesn't change. We're investing in Rook/Ceph to improve your experience, not to increase costs. Storage pricing remains competitive at $0.06/GB-month for SSD storage—the same value proposition you've come to expect from Spot.
Current Status and Future Path
We're approaching this transition thoughtfully:
- February Beta rollout: We’re inviting users to try the new Rook/Ceph offering in Beta, currently available in ORD and IAD
- March Beta rollout: We'll expand Rook/Ceph to each of our larger sites, including SJC, DFW and LON
- April General Availability: Rook/Ceph will become generally available and the recommended storage offering in Spot (subject to user feedback)
- Continued Availability of Cinder CSI: For the foreseeable future, both CSI drivers will coexist and remain available
- Subject to Community Feedback: Based on feedback from our users like you, we may consider deprecating Cinder/CSI support in the future
Staying True to Our Vision
At Rackspace Spot, we're building the Kubernetes platform we wished existed: powerful, reliable, and ridiculously affordable. Our investment in Rook/Ceph reinforces that vision by ensuring storage doesn't become a bottleneck—in performance, features, or developer experience.
We built Spot on the principle that Kubernetes should be accessible to everyone, from developers to startups to enterprises. That means not just an affordablejust affordable compute, but a complete, production-ready platform with storage that works the way you expect it to.
Share your Feedback
We want to hear from you. How does this feature address your experience in Spot? What questions do you have about the new offering? Join the discussion on our GitHub Discussions.
Thank you for choosing Rackspace Spot. Here's to better storage, better reliability, and better Kubernetes for everyone.
About Rackspace Spot: Rackspace Spot delivers fully managed Kubernetes clusters at revolutionary prices through an open market auction model. With free control planes, market-based compute pricing, and enterprise-grade features out of the box, we're making Kubernetes accessible to developers and teams of all sizes. Learn more at spot.rackspace.com.
