In part one of this interview, Nutanix president Sudeesh Nair served the last rites on legacy enterprise software. In this concluding part, he expands on how this vision plays out in a world that’s not developer-centric.
Nair sees the final life cycle of the traditional legacy software coming in a short time span, certainly well less than 10 years. This raises questions along the lines of how Nutanix can find ways to package up the fact that, despite the growth of DevOps, the world is not going to be about developers.
Actually, that is looking at it the wrong way round because they're only being developers because they've got some objectives that they have to meet. What is more, someone else is paying for that, and for those people, the job is not to develop applications, it's to achieve productive results with them. I put it to Nair that one thing which needs to happen is that the developer (in inverted commas and heavily underlined) effectively disappears along with the technology. In other words, becomes masked by functionality and intent. His response:
You're absolutely right in the direction. But I'd make one small exception there which is, for example when platforms like the iPhone come out, I think that some development platforms will continue. Customers are not just going to sit there and hug complexity and develop apps to do Dev Ops. They will focus more and more on business apps and so on.
JP Morgan did a study recently, and asked a bunch of CIOs to name the one company they believe will be more valuable in the future because of public cloud. Surprisingly - or not surprisingly - the number one company was Microsoft. So in this year Microsoft has completely pivoted and understood it is a platform company. They have become more valuable for customers. The SAPs of the world will have to move and make changes in the consumption and delivery of their application platforms. They, I think, will pivot into becoming more services-oriented companies.
I think the biggest axe will fall on the merchants of complexity, which will be the legacy infrastructure as well, and they all know that. There is a reason why Michael Dell is consolidating and building in $18 billion organisation to compete against Amazon. I don't think Michael is doing that to say, 'I'm going to take Amazon out'. What he's doing is, if Amazon is going to take all the new things, someone has to manage the complexity and someone has to manage the declining market. Between EMC and Dell they can do that better than HP. That's the switch, and it's a very straight forward business model.
But what I do think is that the velocity of creative destruction is increasing and the cycle is shorter, which means that I do not think that this will take another 10 years before the new 3rd generation platform and cloud native applications become a huge part of every business.
One issue that still sticks out as something of an anomaly with Nutanix is its commitment to providing the option of a hardware appliance on which to run its hyper-converged environment. It is, however, something that Nair appears to be sanguine about, not least because hardware is still not built consistently across every manufacturer and when hardware does fail, the company policy is to not join in any finger pointing:
So until hardware specifications become more stringent and clear for a new world, we will have to deliver end-to-end peace of mind for customers. We can do that multiple ways, with our own appliances, OEM appliances and maybe at some point in time we can also do something like qualify certain models from channel partners – something like a hardware compatibility list. But I don't think you can build a company of lasting value by running away, by pointing fingers at hardware when the customer is down. You have to own it.
For now however, he describes the partnerships with Intel, Dell and Lenovo as an R&D relationship where engineering teams work very closely together with their engineering and platform teams to qualify every single thing before an appliance is shipped out.
He said that Intel is the company’s biggest hardware partner because the chip company wants to know how they can sell for servers into everywhere with a data center. This does, therefore, preclude any current interest in other hardware architectures, particularly ARM:
We only do a few things, and we do them really well. We always knew that hardware is not something you take for granted and you have to make sure that you don't go there half-cocked. If you are able to do it, you do it. Otherwise don't do it. We've been very careful in terms of making sure that we align ourselves with the Intel roadmap in terms of servers.
While this is an understandable position for the near-term future, it is very likely there will come a time when the ARM architecture starts to invade the hyper-converged appliance and micro-server marketplace, and may end up forcing the Nutanix hand.
Nair does see a change coming in the Cloud Service Provider (CSP) marketplace as well, where Amazon Web Services (AWS) is the current big gorilla. I asked if he sees this current position maturing into one based around many more, smaller, more specialised, service/platform-oriented suppliers? He said:
You are actually drawing the right assumptions here, and there are multiple reasons. Safe Harbor in Europe, for example, and there are laws that prevent people from storing the data outside of the country. So can Amazon open up a data center everywhere? Maybe. But it's never going to be across the whole world.
But the barrier for entry for the small guys is the massive volume of scale that AWS has and the technology that they have to manage to maintain people. They can have hundreds of people who are doing nothing but writing code to maintain their infrastructure. And without that, if they go onto old architecture, the cost becomes so prohibitive, that their services would not be competitive in price to Amazon.
We are fully aligned on this. In fact a large portion of our business is already going to service providers to build their infrastructure, and as we do more and more of this we become absolutely...I don't want to say arms dealer, but we believe that there is someone who is going to glue them together so that users can actually work with us as an OEM so to speak.
There is certainly an opportunity for CSPs to partner with Nutanix to build their value on top and customise the whole thing. Their value add would then be both in meeting niche market needs and in providing local services where issues such as data sovereignty are a problem. And because of the Software Defined Data Center architecture underpinning things, the partners get more than just some basic compute resource on which to build their own applications IP.
There does seem to be a high level of inevitability about the Nair view of the future, though for many established businesses, both vendor and user, it may be coming a good deal faster than they are planning. This is much more about the 'why cloud' than the 'how cloud' and quite possibly represents as much of game-changer as the appearance of the first integrated circuit memory devices. Back then, most vendors and users thought it would never replace core memory. History proved them very wrong much faster than they thought possible.
I would have said the natural 'small c' conservatism of the established user community - and their vendor partners - would point to a ten year span for this to happen. But I certainly would not bet against Nair being right.