In OpenStack Gazpacho (2026.1): Enterprise Grade Storage Advances with NetApp – Part 1 of 2, I wrote about what is already baked into the upstream Gazpacho release with the NetApp Manila drivers. If you haven’t read Part-1, I’d recommend reading it over once because file services are inherently part of any complete cloud service catalog, and you’ll likely find it a compelling argument why NetApp ONTAP should be the backbone for said services.
As OpenStack and the Cinder Project rapidly evolve in response to disrupted market conditions, NetApp continues to deliver exciting new features for VM storage provisioning, protection, and lifecycle management. Management of VM storage is a critical capability for top enterprises and service providers, with a genuine need to solve real-world business problems.
Now, as we dive into Cinder updates in Part-2 of the blog series, we begin to see things diverge just a bit. The Cinder features we are about to discuss are not currently available in the main OpenStack release, but they are publicly available. That means that if you want to take advantage of any of these features, you simply need to reach out to your OpenStack vendor of choice, such as (alphabetically listed) Canonical, Platform9, Red Hat, and others, and let them know you need them to pull in these changes.
With that said, let's start again with a quick review of some terminology:
- Cinder provides a framework for integrating and automating the dynamic provisioning of block devices for VMs. We call these block devices because that’s how the VM sees them, not that they necessarily need to be block devices. They can use SAN protocols, creating iSCSI or FC-connected LUNs or NVMe/TCP-connected namespaces on demand for each VM disk, or NFS file shares to store VMDK or QCOW2-format disks.
- Manila, on the other hand, provides secure file shares for tenants who need enterprise-grade file services.
These Cinder developments focus on the following core pillars:
Let's get into it…
NetApp Cinder: Advancing Block Storage for Mission-Critical OpenStack
Block storage remains the foundation for many mission-critical OpenStack workloads, including databases, ERP systems, and large virtualization environments. In Cinder, the NetApp driver takes a significant step forward by expanding replication, availability, and provisioning efficiency.
Active-Active and Consistency Group Aware Replication
A central theme of the work we’ve been doing in Cinder is stronger, application-aware replication for block workloads. With that in mind, the NetApp Cinder driver is now able to support more advanced replication capabilities. These include:
Zero RPO and near-zero RTO with Symmetric Active-Active SnapMirror active-sync (SM-as)
SnapMirror Asynchronous (SM-a) with NetApp AFF, FAS, and certain ASA systems (those with three-digit numbers, e.g. ASA A400), as well as legacy versions of SM-as, can now be protected as a single CG for the entire backend, keeping write-order consistency across pools on the backend. They can also still be protected as before, where each pool volume is its own replication unit.
Additionally, SM-as, since ONTAP 9.17.1, has supported a cloud-based mediator, so you no longer need to stand up a mediator at a neutral third site to ensure transparent failover.
Asynchronous replication using a single consistency group (CG)
CGs let you protect and failover/failback multiple volumes at the same time, keeping everything in order when something goes wrong at the site level. This is very important for apps that use multiple volumes or need everything to work together smoothly across different data sets. It helps ensure that all file system changes are made in the right order, even if something unexpected happens. This is known as write-order-consistency.
CGs are required with all versions of SM-as, enabling zero-RPO with near-zero RTO for mission-critical applications - with write-order consistency across the dataset.
By integrating OpenStack workflows with ONTAP consistency groups and replication primitives, NetApp enables customers to deploy active-active or metro-style OpenStack architectures with reduced operational complexity and improved resiliency.
Put plainly, OpenStack with the NetApp Cinder driver delivers improved business continuity in the event of a partial or complete outage affecting a rack, room, or even an entire site, commonly referred to as a failure domain.
Learn about ONTAP SnapMirror active sync
Learn about ONTAP SnapMirror asynchronous disaster recovery
Learn about ONTAP consistency groups
Faster VM Provisioning with Cross-Pool Image Caching
In large-scale OpenStack deployments, getting VMs up and running can slow down, especially when you have to store copies of images in lots of different pools.
NetApp has introduced cross-pool Glance image caching for the NetApp Cinder driver. This enhancement allows cached images to be reused across FlexVol pools, resulting in:
- Much faster boot volume creation at scale using storage-native cloning technology
- Significantly improved provisioning performance during large-scale VM deployments
This feature is extremely useful for companies that frequently spin up many virtual machines using standard templates. Even tiny improvements in efficiency can make a big difference in how businesses operate their cloud services. Before NetApp made these changes, customers used to work around the time-to-power-on issue by caching their images; they created a temporary copy of the boot volume on each pool’s FlexVol, enabling intra-volume rapid cloning. Thanks to these updates with OpenStack Cinder, that is no longer required.
One potential downside to consider is that, since pools can reside in different aggregates, there may be reduced image deduplication across aggregates. However, storage efficiency within an individual pool will typically improve.
Improved Day-2 Stability and Reliability
Cinder not only brings new features, but also a bunch of fixes that make block storage more stable and reliable after the initial setup. These improvements include several key enhancements, such as improved handling of certain issues, which overall lead to a more stable day-to-day operation.
One of the most notable enhancements that we made in Cinder was optimizing how we poll volume statistics in ONTAP. These changes have massively improved the performance and reliability of Cinder volume workflows operating at cloud scale.
Some of the other improvements that we made were:
- Snapshot lifecycle handling
- Clone and deletion workflows
- Metadata cleanup
- Service restart behavior
- Pooling reliability and naming consistency
These updates address everyday challenges faced by operations teams when running OpenStack in real-world environments, enhancing its safety and reliability for large-scale deployments.
You can read more about the specific fixes here: 2026.1 Series Release Notes — Cinder Release Notes documentation
Conclusion
The latest OpenStack Gazpacho update, version 2026.1, is exciting news for companies using or considering OpenStack with NetApp ONTAP Manila, and new publicly available features in Cinder bring great new capabilities, like better replication, more control over performance, stronger security, and improved stability.
If your company is already using or considering OpenStack with ONTAP and needs these new capabilities, you can pull these updates into your own build, or work with your preferred vendor. That’s the great thing about open source. The future is in your hands.
More information can be found here:
OpenStack Shared Filesystems (manila) documentation — manila 22.0.1.dev1 documentation
OpenStack Block Storage (Cinder) documentation — cinder 28.0.0.0rc2.dev3 documentation
And general OpenStack guides can be found here: