The pendulum of IT has always swung between two polar opposites – centralised control of IT operations (think mainframes) and decentralised management via servers and PCs in what is known as client-server architecture.
As more organisations bought into the client-server story, server and storage sprawl started to creep up, leading to the advent of virtualisation software, which helps to consolidate servers and improve efficiency.
In recent years, the rise of hyper-convergence has taken consolidation to a new level by abstracting and embedding storage and networking into the compute tier via software, complete with backup, snapshots, data deduplication and other capabilities.
In an interview with Computer Weekly, Sudheesh Nair, president of hyper-convergence software supplier Nutanix, talks about the future of hyper-convergence, its relation with the cloud, and the company’s challenges in building its business in Asia.
Hyper-convergence seems to be all the rage these days. Do you see it being commoditised at some point?
Hyper-convergence will become commoditised – there’s no doubt about it. But that is not new in our industry. About 10 years ago, when Data Domain was started, deduplication was a business that no one else had and they built on it. Today, deduplication has become a feature for many companies in the data storage business. Commoditisation, however, is happening significantly faster. What used to take 10 years to commoditise could take just two to three years – and that’s good news for customers.
Nutanix’s view is very different. We did not begin this company with the end-game of delivering hyper-converged solutions. Rather, we started with the idea that infrastructure complexity had to be eliminated and made storage invisible. Then, three years ago, we went to the next level and made virtualisation invisible. The next step for us would be to make the cloud invisible. So, in our journey, we are not afraid if people expect hyper-convergence to just work and become commoditised, because our approach and destination are different.
Is that why you are rallying the company around the Enterprise Cloud Platform?
Yes, but there are two things about cloud that people need to internalise. First, cloud is something that lives inside a datacentre owned by the likes of Amazon, Microsoft and Singtel. It’s infrastructure that I rent from someone else. Tomorrow’s cloud will be different – it will not be contained within a single datacentre. This has happened before – mainframes used to sit inside datacentres, accessible only via dumb terminals. But what happened next was the emergence of client-server architecture that eventually disrupted mainframes. Similarly, the capabilities of cloud today will bleed out, be miniaturised and reside in internet of things (IoT) devices, such as cars and drones.
When we think about the Enterprise Cloud Platform, it’s basically the data fabric that starts in the cloud and extends to edge devices, as well as the control fabric that manages, automates and orchestrates workloads between the datacentre and the edge. Second, we think cloud should be powered by applications. Take Apple, for example. It doesn’t sell you the cloud. It sells you unlimited music, but when you listen to that music, you are consuming it through the cloud. That invisibility of the cloud will be the future.
Today, some enterprises still operate their own datacentres and private cloud infrastructure for reasons such as security and data sovereignty. Do you think we will ever come to a point where most companies will rely entirely on the public cloud to run their business?
First, there will always be public cloud companies that will build their own technology, but there are also what we call community cloud providers catering to specific industries and geographies, such as an insurance cloud service provider that complies with insurance regulations in Singapore. Nutanix is focused on community cloud providers that want a cloud out of the box.
Second, think of public cloud suppliers as operating a buffet restaurant where you can have a spread. But not everyone wants to eat the same food, and Amazon is not going to change for those with unique requirements. But that does not mean you should go back to the stone age with a three-tiered architecture, and so on. Our bet is that whether you use the public cloud or private cloud, you can have the same experience.
Sudheesh Nair, Nutanix
Some enterprises might decide that they want a private cloud but not own a datacentre. They may rent a cage that is uniquely theirs in a datacentre for which power, cooling and operations are outsourced. The infrastructure stack may be managed by their team or through their consultants, but it is their own cloud with Amazon-like efficiency. This will happen for a lot of companies.
Nutanix has prided itself on giving customers a choice of hardware through partnerships with various hardware suppliers. Some of these suppliers, such as Dell EMC, have now come up with their own hyper-converged appliances. What is Nutanix’s thinking behind those partnerships?
When it comes to competing, we think offence is the best defence. We don’t drive by looking at the rear-view mirror. We have built this company so far by innovating aggressively and pushing the pace of innovation. The idea of hyper-convergence, where storage comes inside the server, is what others are doing.
Every hyper-converged product outside of Nutanix is predominantly built on VMware. There is nothing wrong with VMware, which is now owned by a server company for the first time. Also, VMware was built before containers and Amazon came along, so it is time to reimagine how virtualised infrastructure should be for a new world of security threats, mobility and cloud. Hypervisors also used to be the centre of a lot of things, such as file systems did.
“Customers know that if they want choice and don’t want to be locked in to VMware, we become the only option”
Sudheesh Nair, Nutanix
But today, no one thinks about file systems any more because they have become invisible and just work. Similarly, hypervisors have to become an invisible, reliable component that I don’t have to think about. Now, that goes against VMware’s views.
We don’t have any incumbency or legacy to protect. If you are HP and Dell EMC, you will want to sell 3Par and Vmax, respectively. For us, if a customer says it wants a hyper-converged solution with a Nutanix-type architecture instead of a three-tiered architecture, it can go with Dell EMC XC, and that’s OK. But we believe incumbency and legacy are anchors that will hold you back. Amazon did not build on legacy – it built an entire stack on a clean slate, and natively for a new world. That kind of clean thinking is required to completely change the game.
There are three things that are unique to Nutanix. First, we have the broadest selection of features and hardware platforms on a web-scale architecture. Second, we deliver choices in containers and hypervisors including VMware, Hyper V, Xen and our own hypervisor. Third, we also build management, orchestration and automation outside of VMware. While most companies have commoditised the server with hypervisors, we have commoditised servers and hypervisors, just like how file systems have been commoditised. Customers know that if they want choice and don’t want to be locked in to VMware, we become the only option.
Read more about hyper-convergence
- Converged and hyper-converged infrastructure are ready-tested ways to put compute and storage into the datacentre. But which should you choose for your deployment?
- The battle to rein in IT complexity in the Asia-Pacific region continues with hyper-converged systems that promise to offer cloud-like benefits for enterprises.
- Red Hat’s hyper-converged infrastructure product bundles OS, hypervisor and storage software components from the open source distributor and is aimed at SMEs and remote office use cases.
- Forrester rates 12 hyper-converged infrastructure providers according to products, company strategy and market presence, with Nutanix, Simplivity and Pivot3 coming out on top.
Why build your own hypervisor, then?
It’s very simple. It’s about user experience. Apple has been able to deliver high performance and a consistent experience because of the tight integration between hardware and software. Google does this by partnering with Samsung. For customers that want complete peace of mind for hardware, hyper-converged software and application support, Nutanix’s NX platform works well.
We also provide customers with a similar experience by working with Dell EMC, Lenovo and IBM, except that hardware support is provided by those OEM partners and we will support the software. The new model we announced was to decouple software from the hardware from HPE and Cisco, which probably don’t want to support Nutanix, although they don’t have a choice because our platform works like any other operating system. So they’ll have to support it, in a sense.
Nutanix Go offers subscription-based pricing for companies looking to build enterprise clouds using operating budgets. How is Nutanix gearing up to support a consumption-based model?
It is easy to take any architecture and wrap it in financial creativity, such as leasing and subscription models. That is good but what is best would be the architecture itself supporting a consumption-based model. That means you should be able to take a three-node system, turn it into a 300-node system and back into a 200-node system. This requires metadata to be truly elastic and scalable.
We have the advantage of being truly elastic, so the consumption model is easy to execute, because we can always add a node whenever a customer needs it and we charge accordingly. The problem has always been about the hardware cost and how to keep it in the books. But by announcing software-only products, we are now able to decouple hardware from software, making it much easier to offer subscription pricing.
Nutanix Go is not available in the Asia Pacific region yet, as we still have some tweaking to do. That said, I don’t think all customers will go for a consumption model because it could cost more over time, just like renting. Customers that don’t have a cash flow problem may still prefer capital expenditure and depreciate their assets over time because that is better for their books compared to leasing.
What are your thoughts about the adoption of hyper-convergence in Asia-Pacific, a diverse region with differing levels of technology maturity?
Nutanix has always been very aware of international markets. We have striven to be a global company since our early days, but it takes some time for Asian customers to trust young US companies. They have seen one too many young companies come and go.
We have done a pretty good job of methodically and patiently investing and building our business in the region. About four years ago, we had no customers here, but today we have major banks and telcos as customers. We are also seeing government agencies spending tens of millions of dollars with Nutanix for the first time. Customers don’t trust you just because your technology is good. They want to see that your company has ambition, and that its people, support and roadmap are good.
Most of our challenges in Asia have been about finding the right people and retaining them. And in Asia, unlike the US and Europe, every country is like a different planet. In ASEAN alone, there are at least six different languages and 10 different cultures. For a young company, it was very hard to invest in the region, especially in identifying people and channel partners that can cover the whole region. In the US, you can hire someone in Boston to cover the eastern region. Here, you have to hire different people to cover different countries. But now that we have grown and matured, we are aggressively investing in the region.
Some technology companies have delisted from the markets in a bid to refocus their company around longer term growth, rather than quarterly investor expectations. What are your thoughts on that, now that Nutanix has become a public company?
There are arguments for both sides. Going public brings discipline to a company from an investment and ROI [return on investment] point of view. However, public markets tend to force you to look at tactical instead of strategic things. What we have done is to put in place a dual-class share structure as part of our IPO [initial public offering]. That gives insiders a bit of an edge when it comes to navigating strategy.
We have also selected investors who know that this is a long game. Public investors are also smart enough to know whether a technology has the scope or TAM [total addressable market] to expand on a long-term strategy. So, if you’re operating in the backup space, it’s ridiculous to expect to become a $20bn company.
In our case, if we continue to execute, you could make an argument that the TAM is hundreds of billions of dollars, because it’s not just about compute, storage and network, it’s also hybrid cloud and IoT, which we’re investing in.