Paid feature The promise of hyperconverged infrastructure (HCI) is that companies get an easy to manage, highly efficient resource, where scaling up or down is simply a matter of adding or subtracting more commodity servers.
No need to disappear down the rabbit hole of separate silos for storage, compute and other physical infrastructure.
The appeal for huge enterprises – and the huge companies that want to supply them – is clear. But what about the smaller guys? There are plenty of organizations that do not fit the mainstream HCI suppliers’ customer “personas” and have quite specific requirements of their own yet still want to leverage the benefits of HCI. So says Andrew Tuzov, director of business development at StarWind Software.
StarWind has had plenty of time to come to this conclusion, with its roots stretching back nearly 18 years, when it was founded by COO Artem Berman and CEO, CTO and chief architect, Anton Kolomeytsev. Veeam founders Ratmir Timashev and Andrei Baronov subsequently invested in the firm and sit on the StarWind board.
The company initially produced iSCSI SAN software, before developing its VSAN HCI software from 2008. About five years ago, says Tuzov, “we realised that while we were a software company, most of our customers were buying the software and then having to build their own hardware.”
Many of these customers were small or medium-sized organizations. He adds that such companies were often over-focused on current technology requirements, which can be a constraint as they grow. So, it made sense to build a complete solution for them.
More broadly, the big players in the HCI market were, and are, pulling in different directions, Tuzov claims. On the one hand, the likes of Microsoft are working to ultimately bring people into the cloud. “The Azure HCI stack is kind of a play to bring more people into the hybrid cloud space to facilitate an organic transition.”
But, he continues, for small and medium businesses, cloud is not always the best option, whether because of concerns about data control and sovereignty, or the fact that even the biggest cloud falls over on occasion.
At the same time,more and more data is generated “at the periphery” thanks to IoT, smart vehicles, and other applications. “Sending this data to primary data centers directly does not really make sense,” Tuzov says. “It's much better to structure and work with the data at the edge, and then have that structured and organised data sent to the primary data centres.”
He argues the focus for the mainstream HCI players is almost always on the enterprise. When they do think about small and medium customers, their response is to offer something that is out of the SME price range or a drastically cut-down version of their enterprise product.
Enterprise grade for the rest of us
By leveraging its inhouse development team, StarWind was able to build the all-flash StarWind HyperConverged Appliance (HCA), “an enterprise grade product which is aimed at SMBs.”
This is not a one-size-fits-all device: “We treat each deployment as an individual project, where we will completely analyse the technical background, the business background, and we will build a bespoke solution.”
A key differentiator of the StarWind HCA is the way it utilizes SSDs. StarWind’s engineers concluded that read intensive SSDs offered a similar level of performance at a lower cost compared to mixed use SSDs, but with less write endurance.
StarWind “decided to alleviate that by creating a hybrid appliance utilizing mixed use SSDs as the hot tier and then using read intensive SSDs as the cold tier, which allows for much higher performance than traditional hybrid appliances. But at the same time, proper resiliency and a sensible cost without having to pay the all-flash tax.”
The next step, Tuzov says, was working closely with Intel and incorporating the chip giant’s Optane drives, which he describes as “the best storage medium that is available on the market right now, both performance wise and resiliency wise.”
StarWind uses Optane as a cache tier in its HCA, with read-intensive drives acting as the underlying storage, he explains, “Which takes all the benefits we already had with the combination of mixed use and read intensive, but just takes it that much further.”
StarWind’s figures show the combination of its architecture and software allows effective utilization of the cluster's resources, with six 9s availability and highly effective storage utilization and IOPs processing. The HCA can be custom-built to meet your requirements both in terms of compute power and storage.
StarWind BA: NVMe changes the game
StarWind recently added a renewed backup solution to its model line – the StarWind Backup Appliance (BA).
“We have a fantastic solution that offers really high uptime for the applications of the customers, but we decided to complement it with a backup solution, which just completely rounds everything off,” says Tuzov. “We can make sure that even if your primary infrastructure fails, you still have ways to recover and recover fast.”
StarWind BA uses NVMe drives, which completely changes the traditional backup paradigm, according to Tuzov.
“Usually when we talk about backups, it's all about the concepts of RTO and RPO. It's how fast you can recover data and how much data you're going to lose if you recover from a backup. But an NVMe backup appliance means that your recoveries are pretty much near instant.” This removes the traditional compromise between performance, data and uptime, he says. “It’s pretty much an extension of your production infrastructure.”
Fit for purpose
StarWind continues to offer the bespoke approach, delivered by its engineering and service teams in Ukraine, as it broadens its hardware appliance portfolio, Tuzov says. “The way we open discussions is usually actually talking about [customers’] infrastructure, about the problems they're facing, about the challenges they want to overcome.”
That includes how the customer needs to compete within their specific industry ... “because each industry is extremely competitive. Nowadays, we're living in a time where there are very few to none blue oceans, and everything is just a red ocean of competition.”
From there, the conversation will lead to “we've faced the same issues that you faced with these types of customers …. So here are a number of approaches you could take.. Now, the benefits to each one are this, the downsides to each are this. And then we work with them on identifying which approach is the right fit for them.”
This customer-centric perspective has earned StarWind one of the highest customer satisfaction scores in Gartner’s Magic Quadrant for Hyperconverged Infrastructure Software, and recognition for “innovation in the hardware and software HCI layers, which makes it an attractive price/performance option for SMEs.”
You need to run as fast as you can to stay in the same place
This is more important than ever as storage technology races ahead, according to StarWind. The next frontier is working on standards around NVMe over Fabric. And although StarWind does not open source its software, the company is working with the key industry groups and vendors around the future of NVMe-oF. The key challenge is less about the hardware itself, which will become increasingly affordable, but ensuring the software can keep up.
“We’ve gotten to a point where the hardware is so fast, but the software is having a problem efficiently utilizing it. And we are trying to make sure that we're going to be one of the vendors that is going to allow people to efficiently utilize more NVMe arrays and get the most out of the underlying storage.”
This means developing software for the hardware that’s on the roadmap in one to two years time, and the workload that customers will want to run on it – even if the customer doesn’t yet know what those workloads are likely to be.
“It's a bit like Lewis Carroll with Alice Through the Looking Glass,” Tuzov says. “Where you need to run as fast as you can to stay in the same place.”
This article is sponsored by StarWind Software.