UK managed service provider (MSP) Inca Cloud has chosen Nebulon cloud-defined storage over hyper-converged infrastructure (HCI), claiming a 40% saving by avoiding overheads inherent in HCI.
Meanwhile, Nebulon announced smartIaaS (infrastructure as a service) while also publicising early access UK customer, Inca Cloud, which will use it to support its new WSO by Inca hybrid cloud and multicloud offering. Inca technical director John Thorpe cited Nutanix, in particular, as a too costly hyper-converged option it didn’t choose for these projects.
With WSO, Inca will provide enterprise clients with a single platform to provision and maintain workloads in the public, hosted or private cloud datacentre. Inca chose Nebulon as storage for customer data it will host. The service also provides an extensive library of machine images, as well as a common cloud-like experience independent of datacentre type.
“We are targeting customers that have on-premise infrastructure but want to extend that to the cloud,” said Thorpe. “We want to help customers remove the complexity of multicloud, where there can be different interfaces, different skillsets and requirements, different overheads, and so on.”
Thorpe said customers would typically use hybrid cloud or multicloud on specific elements of a workload, such as bursting a web front end to the cloud to keep up with demand for payment, while keeping data in the on-premise or hosted environment.
“Also, different public clouds have different use cases for organisations, so for example a Microsoft shop is likely to want to use Azure because they’d be more supported,” said Thorpe.
Inca has started the roll-out with three server instances based on SuperMicro hardware and running Nebulon. These will provide storage for customers and will form storage instances, manageable by application programming interface (API), to form a “logically private cloud”, said Thorpe.
With Nebulon, physical storage is in commodity hardware but with PCIe slot-resident storage processing units (SPUs).
These offload storage input/output (I/O) processing and management from the server to the Nebulon hardware.
SPUs replace RAID cards and storage host bus adapters in servers. Each SPU has Gigabit Ethernet (GbE) ports that form the data plane, along with 1GbE cloud connectivity.
Local PCIe-connected SPUs connect flash storage in the node and emulates storage controller functionality. Provisioning and application, server and storage metrics are offloaded to the Nebulon ON cloud and are accessible via browser to Inca, in this case, but not their customers.
A Nebulon cluster, known as an nPod, can scale to 32 servers with data services configured in the Nebulon ON cloud portal that include mirrors, snapshots and volumes.
Thorpe said Nebulon for Inca was “a way of providing storage in an HCI environment”, but avoiding Nutanix, which he said had become too costly.
“Nutanix has become its own ecosystem. It has VMware-d and started to consume other products,” he said. “Nebulon is more agnostic. And the benefit is that it doesn’t have the cost overhead of other HCI products that have become part of a bigger ecosystem. Then, for example, there is the fact that Nutanix has a controller VM [virtual machine] that runs servers and this creates an overhead. Nutanix has done a lot of good work to mitigate this, but it’s still an overhead.”
This lack of controller overhead is a key attraction of Nebulon. According to Thorpe, this translates to reduced server requirements of around 40%. “You can do with four servers what you previously needed six for to get the same capabilities. At scale, that can mean a lot, and we can pass it on to customers.”
