Cloud-Defined Storage

Increase asset utilization, operational simplicity, and agility while taking advantage of cloud innovation.



I had the opportunity to meet with Siamak Nazari, CEO, Craig Nunes, COO, and Tobias Flitsch, Head of Product at Nebulon during the 35th IT Press Tour.


Nebulon is coming out of stealth-mode to introduce the concept of cloud-defined storage and address the CIO challenges of 1) increasing asset utilization while reducing cost and infrastructure footprint; 2) increasing operational simplicity and agility while reducing operational overhead and accelerating the deployment of new services; and, 3) taking advantage of cloud innovation across the enterprise and giving application owners a self-service, on-premise experience.


According to the Nebulon team, cloud-defined storage is on-premise, server-based, enterprise-class storage that consumes no server CPU or memory resources since it is managed through the cloud. The secure, AI-driven, cloud-based control plane is called Nebulon ON. The application server clusters with PCIe-based services processing units (SPUs) replacing the RAID or FC adapters.


Ideal customers of the solution are infrastructure managers and application owners with a need for low-latency, mission-critical workloads on-premises. They’ve tried SDS or HCI (hyper-converged infrastructure), but SLA or other workload restrictions forced continued use of external storage arrays. They’re managing 10’s, 100’s, or 1000’s of servers and may have multiple sites or edge locations. Example customers include cloud service providers (CSPs) looking for cost-effective ways to deliver value-added services to clients and mid-to-large enterprises under pressure to reduce costs and offload the overhead on operations staff.


Customers will buy Nebulon from their existing server vendor. There is no need to add another vendor. SPU subscription is available as a server option from HPE, Supermicro and more server vendors are being added. All quotes, order fulfillment, and support are provided by server vendors.


SPU replaces the storage controller in the application server and runs in any 2u/24 drive server, connects to the server SSDS like any RAID card, presents local or shared volumes to the application. A group of SPU-equipped servers is an nPod.


A server with an SPU will have all the attributes of an enterprise all-flash-array (AFA) within a card: hypervisor/OS agnostic, enterprise data services, AFA latencies, 1.33X, petabyte-scale nPods, half the cost of expensive external arrays, and workload awareness using tailored application templates.


CDS differs from HCI in the following ways: 1) CDS uses no server resources while HCI uses 20 - 25%. This results in fewer servers being bought and lower licensing costs. 2) CDS supports any hypervisors or operating systems while HCI is typically VMware-only. This provides deployment flexibility. 3) With CDS, the fault domain is server independent while it’s server-dependent with HCI. This means local storage is available even when the storage is down. 4) CDS’s write cache is NVram while HCI’s is write-intensive SSD. 5) CDS’ SSD firmware management is automated while HCI has none. 6) The CDS control plane is cloud-based while the HCI control plane is cluster-based.


Management from the cloud control plane means it’s always up-to-date and available as-a-service. Users can make use of new features instantly and it’s enriched with AIOps capabilities. The SPU offers a lightweight storage operating system with fast and infrequent updates with no disruption and easy to schedule updates.


There is secure management of the customer’s global infrastructure. It’s disaster tolerant with multiple availability zones so management is always available for any infrastructure scale. There are zero operations impact is cloud connectivity is lost. Role-based access control ensures all actions are secured and audited to allow for and monitor delegated management. Data is encrypted in-flight and at-rest for user interactions, SPU’s, and cloud service to keep all data private. Security tokens are used so only authorized users within the firewall can make changes.


CDS benefits include: 1) global infrastructure insights with 10’s of 1000’s of server, storage, and app metrics every hour and versatile reporting for informed decision-making; 2)AI-assisted administration with AIOps for predictive maintenance and faster troubleshooting and zero-touch provisioning; 3) API-driven automation to control all devices from a single URL in the cloud and automated software updates and fleet management.


CDS offers self-service infrastructure provisioning using app templates. Curated, reusable definitions of server-storage configuration for specific applications like VMware, Kubernetes (K8s), MongoDB, and more. Repeatable, consistent infrastructure deployed at scale. Template automation and constant optimization using system telemetry.


Key Takeaways

  • Avoid the expense of array-based storage without the restrictions of SDS and HCI.

  • Enable self-service for application owners while reducing the burden on infrastructure managers.

  • Available from your existing server vendor so there’s no need to add a new vendor.


Drop Me a Line, Let Me Know What You Think

© 2020 by Tom Smith | ctsmithiii@gmail.com | @ctsmithiii