SoftIron to start selling 'fully turnkey' cloud platform

Hypercloud will be facing some stiff competition in the hyperconverged infrastructure space, say analysts

Storage player SoftIron is extending its portfolio to the rest of the datacenter with the launch of a complete cloud infrastructure platform - a bold effort to take on better known brands in their own back yard.

This is all paid via a consumption model, with everything said to be included in the purchase price and with no additional licensing required to access any features.

SoftIron, best known for its HyperDrive Ceph-based high-performance storage arrays, claims its HyperCloud Intelligent Cloud Fabric removes the complexity of building and managing a private or hybrid cloud because its hardware and software have been designed from scratch to work together.

Cloud computing is being widely adopted because it offers an easier way to consume resources, often through a self-service front-end. But this hides a mass of complexity underneath for the IT team to manage, according to SoftIron chief marketing officer Andrew Moloney.

“When you look under the hood, and look at how clouds are built today, they are a myriad of different layers and technologies, all of which have to be threaded together in various ways,” he said.

This can include hardware from a variety of vendors, the layers of networking, and the multiple layers of software that are required to pull it all together and present it as a single pool of resources.

The only people that have really solved all of that complexity when it comes to delivering a private cloud or hybrid cloud is AWS with Outposts and Microsoft with Azure Stack Hub, according to Moloney, “but they're only able to do that through brute force engineering, the fact that they have tens of thousands of engineers, but at the expense, for customers, that you're locked into that ecosystem.”

In contrast, HyperCloud enables any organization to build and operate their own cloud, even if their IT team doesn’t have cloud specialists, Moloney claims.

“We're talking about the ability to build a completely resilient, fully consumable private cloud in 8U of rack space in less than half a day, and the level of configuration required is an absolute minimum,” he said.

So what does HyperCloud comprise? The basic building blocks are discrete Compute, Storage and Interconnect nodes, unlike hyperconverged infrastructure (HCI) that integrates all the hardware into a single node.

This was a very intentional decision, according to SoftIron chief technology officer Kenny van Alstyne, so that compute and storage resources can be scaled separately, as required.

“And those architectures [HCI] generally have an upper limit, they're not really designed for building full-scale cloud infrastructure, which is what we're showing here,” he said.

With HyperCloud, “you buy as much Fabric Interconnect, Compute and Storage as you need. And as you need more, you can buy that incrementally,” Moloney said. There are of course, ongoing support costs on top of the purchase price.

“In the same way that you buy hardware today, you buy the product and you buy maintenance support as required, but the key thing is, unlike a lot of those other ways of building a cloud, you've got complete access to all of it. You've got complete control over all of it and access to all of its features and capability out of the box,” he added.

The nodes themselves are stateless, apart from the distributed control plane that lives in the Interconnect nodes, which hold all of the cluster states, including the boot images to boot all of the nodes, and the cluster configuration such as customer management, and IP address information, Van Alstyne said.

SoftIron told us that the HyperCloud nodes are based on the same hardware design as its existing products, but have different configurations, including no boot drives as they are "stateless", and they have different memory options. The nodes are either X86 or ARM based, depending on which is the optimal choice. At launch, compute nodes are available in 8 and 16 core (16 and 32 thread) options, and storage nodes are available in a range of HDD, SSD and NVMe options, which the company said enables a broad set of use cases to be accommodated in any HyperCloud fabric.

The OS that runs on the nodes has the ability to handle virtual machines, containers, or bare metal server instances, and can deliver file, block and object storage all within the same platform, according to SoftIron, with all the automation required to allow users to consume those resources.

This includes multi-tenancy support right out of the box, the company said, and allows SoftIron to target cloud providers as well as organizations wanting a private cloud.

“We're already seeing a lot of interest in this from the second tier of cloud service provider or managed service provider who wants to build a managed service that's either focused in a region or focused on a specific market or a use case,” said Moloney, “But they want to do that in a way which is profitable and manageable, if you like without having to hire an army of specialists.” SoftIron told us that its OS is Linux-based, and that the hypervisor is based on KVM, but the company maintains its own fork of the source code.

“At launch, we’ll have integration with some of some of the more popular container technologies like LXC, Docker and Kubernetes, and some of those can be deployed on bare metal, and some can be deployed inside other containers or inside virtual machines,” Moloney said.

There are also APIs so that HyperCloud can integrate with automation or infrastructure-as-code tools like TerraForm, Ansible, Chef and Puppet.

Another key capability is linking with public clouds, as part of a hybrid cloud strategy, and HyperCloud will integrate with AWS and Azure at launch, enabling users to expand out and use resources from these two clouds.

In the case of AWS, this works by having an AMI (Amazon Machine Image) that maps to one of the internal images on your local HyperCloud storage, and when a workload gets deployed, it will reference that AMI ID to create the machine instance, SoftIron told us.

On the face of it, delivering a complete turnkey cloud stack is a very ambitious move for any company, and many organizations may be hesitant to invest in what is a proprietary solution from a relative newcomer in the market.

“The SoftIron plans are ambitious and they will be facing some stiff competition in the Hyper-Converged Infrastructure space such as HPE, Dell, Lenovo and Cisco to name just a few of the big players,” said Futurum Research senior analyst Steven Dickens.

However, many organizations are desperate for simplicity and to be able to focus on business value-generating activities rather than managing hardware, he added.

“The SoftIron approach leans into this market demand. Many customers are making strategic choices for hybrid cloud infrastructure, and with this announcement SoftIron is should be on the list of vendors to be considered.”

IDC research director Chris Kanaracus warned that HyperCloud is a version 1.0 product and thus unproven, but said SoftIron has a strong base of experience building, selling and servicing its Ceph-based storage appliances.

“HyperCloud is obviously a very ambitious gambit for them as it adds so much functionality to the software stack, for example VMs and containers in the same system. Not only that, but it supports both x86 and Arm workloads at launch, and they intend to add bare metal soon. I’ll be keen to see evidence of customer success in production at meaningful scale with HyperCloud,” he said.

Kanaracus also said that SoftIron’s approach to scaling out is intriguing.

“The systems don’t have any identity or persistent configuration information loaded onto them. That info is held within the control plane layer, which they’ve distributed into the network. The upshot is that if a HyperCloud system fails, you can call them, they’ll ship a replacement and it plugs in, starts running. The analogy is that you can treat HyperCloud nodes like cattle, not pets,” he explained.

SoftIron told us that the HyperCloud Intelligent Cloud Fabric will be ready to order from launch, with standard delivery times applicable, and it will be available via the company’s existing channel partners. ®

Similar topics


Other stories you might like

Biting the hand that feeds IT © 1998–2022