What Nutanix Does

I embarrassed myself the last time I asked a question about Nutanix on this board, and took Saul’s words about “doing the work” to heart. I remember seeing a link to the NTNX Bible (huge tip to CMFALieberman) a few weeks ago, which is the company blueprint. In attempts to better understand what Nutanix actually does, I went ahead and read it. At over 300 pages it is incredibly dense, and I’ve only included notes regarding the company’s thesis, not the hundreds and hundreds of pages detailing how the software works button by button, and coding line by coding line.

I’ve included some backstory about mainframe computers and kernel space, as such particular information was necessary to absorb before truly grasping the necessary ins and outs of the company. My apologies for the length, and if this has already been covered on this board. I know Nutanix has been heavily discussed here, but I couldn’t find something similar to what I’ve done below.

A Quick History on “The Cloud"

  • In the beginning of computers, all data lived on a large mainframe. The only way to access the data was to use that large machine, and only the person using that machine had access to the information. This created a “silo”, or information walled off from other users/machines.

  • Over time, access to the data got smaller (laptops) and more portable (external hard drives/phones). However, only the person using that device had access to the information. Silos still remained.

  • Data is now moving to the cloud, which allows users to share and access that data any location on Earth, provided they have an internet connection. This removes silos, and is being adapted by companies all across the globe.

Pillars

The core pillars of any cloud service are:

  • Self Service (the ability for a person to easily access the data)
  • Service Level Agreements (SLA) (this guarantees that the cloud remains active and operational, and at a certain speed)
  • Fractional Consumption Models (while some services are free, such as say Gmail, other services can be billed out to the customer)

Cloud Classifications

The last pillar above (ie, the money making one) can subsequently be broken down into three classifications. These are:

  • Software as a Service (Microsoft Word)
  • Platform as a Service (Apple’s App Store)
  • Infrastructure as a Service (this would be the brain and central nervous system, keeping all things in your company working properly).

Cloud Latency
Faster is better, especially on the Cloud. The more users accessing the network, the more memory (RAM) the server must posses to avoid latency. Servers alleviate latency two different ways:

Kernel Space

  • The most priviliged part of the Operating System (OS)
  • Handles memory management
  • Contains the physical device drivers

User Space

  • Basically “everything else”
  • Handles things such as opening Microsoft Word

When the user requests something done (ie, save the latest presentation on Power Point), the system routes that request to either the Kernel Space or the User Space, and the task is processed. This of course takes memory (or lets also call it computational power) away from other programs also running, thus latency is introduced. The more complicated the request, the more latency. The more requests happening at the same time, the more latency. This is why the first solution for a computer running slow is to “close all other programs” as those programs might all be sending requests to the kernel/user space, thus depleting the overall computational power of the device.

A simple flow path of this is as follows:

I WANT TO DO SOMETHING > COMPUTER SENDS THAT MESSAGE TO THE KERNEL > THE KERNEL STOPS WHAT IT IS DOING > THE KERNEL EXECUTES YOUR DESIRE > YOU GET THE RESULT

These requests can be handled in two separate ways:

Polling

  • Constantly asking the kernel/user space if it needs anything
  • Requires constant CPU, but much lower latency

Interrupt

  • Only acts when given a direction
  • Tells Kernel to stop what it is doing and execute your tasks
  • Much higher latency

As a result of this, Operating Systems are being written more towards Polling than Interrupting, which results in this flow path:

I WANT TO DO SOMETHING > COMPUTER WAS ALREADY CONSTANTLY CHECKING IF YOU WANTED TO DO SOMETHING, AND HOLDING THAT MEMORY TO EXECUTE IT > YOU GET THE RESULT

In even simpler terms, kernel space is the CEO and doesn’t need to be bothered with high volumes of mundane tasks that lower level employees (User Space) can handle. This frees up the CEO to put all their efforts towards solving the biggest problems of the day.

All that said, we are now getting into WHAT Nutanix does.

Hyper-Convergence

Boiled down, hyper-convergence is natively combining two or more separate components (say the graphics card and the cooling fan motor) into a single unity thus reducing computational power required. For example, when the graphics card on your computer goes into overdrive to display a large visual element, the onboard cooling fan motor natively speeds up and better cools the system core. Taken directly from the bible, “In the case of Nutanix, we natively converge computational tasks to form a single node used in our application.”

Software-Defined Intelligence

Taken directly from the bible, “Software-defined intelligence is taking the core logic (ie, its central nervous system) away from specialized hardware (ie, an application specific chip such as one that manages only the microphone in your computer) and doing it in software in commodity hardware (widely available, not specialized). At Nutanix, we take traditional logic and put that into software that runs on standard hardware.”

In layman’s terms, Nutanix software does what complicated hardware used to do on widely accessible and affordable hardware.

The benefits of this to a business are as follows:

  • Rapid release cycles (software can be updated constantly and instantly, instead of waiting for new hardware product cycles)

  • Eliminates the need for fancy, specialized (ie, expensive) hardware

  • Lifespan investment protection (ie, you can run new software on old hardware. Think about how you are able to update your computer operating system without changing out hardware components).

Distributed Autonomous Systems

Taken directly from the bible, “distributed autonomous systems involve moving away from the traditional concept of having a single unit (ie, motherboard or graphics card) responsible for doing something and distributing that role among all nodes within the cluster. Traditionally, vendors have assumed that hardware will be reliable, which, in most cases is true. However, hardware will eventually fail and handling that fault in an elegant and non-disruptive way is key. These distributed systems are designed to accommodate and remediate failure, to form something that is self-healing and autonomous. In the event of a component (ie, hardware) failure, the system will transparently handle and remediate the failure, continuing to operate as expected.”

In layman’s terms, hardware failure is a huge pain in the ass so why not use software that is cheaper, more scalable, redundant and consistently checks up on, and fixes, itself automatically?

Incremental and Linear Scale Out

Taken directly from the bible, “Incremental and linear scale out relates to the ability to start with a certain set of resources and, as needed, scale them out while linearly increasing the performance of the system. Traditionally [with hardware], you’d have 3-layers of components: servers, storage, and network – all of which are scaled independently. As an example, when you scale out the number of servers you’re not necessarily scaling out your storage performance. With a hyper-converged platform like Nutanix, however, you have the ability to improve all three at the same time.”

In layman’s terms, think about this like a traditional gas engine versus the software running a Tesla. Swapping a part on a gas engine does not mean that all other pre-existing parts of that engine are now running a peak capacity. On a Tesla, however, when you update the software, you optimize the entire operational system of the vehicle.

Hopefully all of the above better defines what Nutanix does. It was helpful for me.

Brandon

134 Likes

Brandon,

Thanks for the kind words and for taking the time to read the Nutanix Bible and write up such a great summary. Really helpful and well constructed.

I looks forward to reading more of your posts.

1 Like

Brandon, great job writing this up!
Thank you for your time and effort. It’s very much appreciated.

Speedy

2 Likes

Brandon, that was great. You made a complicated and difficult set of concepts understandable (well, pretty understandable, anyway! :grinning:)

You’ve com a long way in a short time. Thanks for tackling that. I never would have dared.

Saul

7 Likes

****ing incredible. Thank you, Brandon!

Bear

3 Likes

Wow Brandon, great job. You took “doing the work” to another level. There are many many posts here trying to describe exactly what Nutanix does, and I was able to read yours start to finish without my eyes forming an impenetrable glaze over them.

Thank you,
Matt (Techie simpleton) :slight_smile:

7 Likes

@brandonwahl

Could you provide a link to the “NTNX Bible” document you referred to?

I think it’s admirable that you undertook understanding the company better, but I think I can offer you a couple of clarifications around your understand of various *aaS levels, the nature of kernel and userspace, the nature of asynchrous (poll vs interupt) processing, where the term hyper-converged infrastructure comes from and what it means, and what benefits are unique to it.

I am trying to help and would like to read the same doc before offering any feedback. I think that Nutanix takes advantage of jargon to market their software and company in the most impressive sounding way possible. While I have no NTNX position of my own (long or short) it’s clear their customers are willing to spend money on them and maybe they’ll be a big equity winner. But understanding how the tech maps to ROI at a CIO or CTO purchasing level in their customers is probably important.

3 Likes

But understanding how the tech maps to ROI at a CIO or CTO purchasing level in their customers is probably important.

Well put. Btw, here it is. I just googled “Nutanix Bible.” http://nutanixbible.com/

Also, while we know they had x number of 1m+ lifetime contracts last quarter and x number of large customers…as far as I remember, I have never seen a retention rate. Has anyone?

And also, do we know how pricing works? Is it ratable? Based on seats? ???

Bear

2 Likes

HCI is very simple, it takes compute, storage, and networking away from separate (and expensive) proprietary hardware and software and enables the same functions to work on commodity hardware with the HCI operating system (as that is basically what it is) operating all three of them through one control panel.

It enables plug and play of any other commodity hardware that otherwise works with the chosen HCI system, it creates fail safes if one piece of hardware breaks, and it creates (when used in the correct environment) large savings in ROI. The ROI is not only from savings in equipment but also in maintenance and labor costs (something like Arista did to Cisco).

HCI is not the best solution. But when it is a good fit it enables an enterprise to quickly create their own datacenter from scratch (as an example) and then scale it with minimal (compared to previous processes) labor and maintenance.

HCI has largely not touched the largest of data centers to date (although that is starting to change as larger and larger deployments are now being made), and HCI is not always the best solution where something like what Pure is trying to offer, direct attached storage (modernized with faster networking capabilties) are used (as are often used in hyper scale data centers and in AI set ups as an example).

With the network expanding to the edge (IoT) HCI will enable the edge to become part of the data center and controlled by the single interface.

Sure, there are many details. But who cares. There are also many details as to how Windows works vis a vie Linux or Apple OS. So what. The OS systems themselves handle the complexities, and have their quirks, and their strengths and weaknesses. It does not do me any good as an investor to know that Windows may have a better read write on high level Excel calculations vs. Apple’s superiority when paired with x, y, or z functionality.

There is no disruptive technology anywhere near to market (although one of the founders of Nutanix has created and has a hot pre-IPO business that takes HCI to a different level (just storage I guess, I have not looked into it in more detail), but that is not going to displace Nutanix or VMWare HCI.

The technology obviously creates superior ROI that heretofore was not reasonably possible on a mass scale of customers, the technology is obviously moving up-scale as larger and larger deals are struct and at faster and faster rates, the whose who of anyone in that particular industry is trying to offer their own version of HCI, and the facts show, disputably, that Nutanix and VMWare are not only leading the market, but growing their lead in the market.

Now you can either invest after everyone who is going to use the technology adopts it (and thus when growth stops - but there is no risk that HCI is really, really, really for real, or not).

The technicality of how a kernel works is of no value to the investment thesis.

But have away at it.

What I want to know is (1) is it in hyper growth? Yes, (2) is the market important and very large (yes), is the company the market leader (yes, or at worse a duopoly), (3) does the company have sustainable competitive advantages (yes), is there something disruptive that is not just hypothetical but actually doing something we can materially track in the marketplace (no), (5) does the market cap already consume the opportunity here (no.).

Anyone have anything to argue with any of those points?

Tinker

28 Likes

It’s like an IT department for your IT department.

Remember the old Maytag commercials where the Maytag guy sat around day after day with basically nothing to do because the Maytag stuff always worked and never broke? No one ever called him for a service call.

The software that Nutanix provides (sort of) turns your IT guys into Maytag guys.

6 Likes

“The technicality of how a kernel works is of no value to the investment thesis.”

Perhaps you missed it but the person who I was replying to brought up kernels, not me.

Anyway, you know what hyper-converged infrastructure is? There’s an old joke:

“What’s the white stuff in bird poop”

“That’s bird poop, too”

Hyper-converged Infrastructure is Infrastructure. It is not a new category of spending. It is a new-ish category of management software that is consuming a portion of the TAM of existing infrastructure spending. The true competition is pure cloud workloads, which is competing for the same TAM.

Hyper growth is hyper until it’s not, as we recently saw with NVDA. Adoption follows an s-curve, and that curve is getting more and more compressed. Understanding the upper bound is important to understand the runway for hyper growth and the NPV of a given investment.

9 Likes

Net Retention Rate Would not be an appropriate metric for Nutanix. Maybe as they grow the SaaS they can provide that for those.

Because Nutanix Core sales which makes up the majority of sales, is somebody ordering a whole bunch of Nodes and cutting Nutanix a check for the software billed on a per node basis. Or from somebody adding more nodes to an existing system and scaling out. Or now that terms are transitioning to subscription the ratability period cycling. Or from an existing customer renewing a subscription contract or replacing a node stack.

Customer Retention rate is greater than 90%. Can’t be greater than 100%.

If you want to know more about how they count revenue I did a dive into the 10-q and posted this a few weeks ago last time it came up.

https://discussion.fool.com/more-detail-on-the-how-nutanix-earns…
Darth

5 Likes

Darth’s link said:

Portable Software($127M +104% yoy)
-Software entitlement and support subscriptions, Cloud based SaaS offerings, and separately purchased software term-based licenses. The first two represent $83M (+62% yoy)and are recognized ratably over the contract period. The latter ($44M up from $11.3M in year prior) is recognized at the time the software is transferred to customer. The weighted trailing 4 quarter average is 3.6 years.

If we’re dealing with 127m/quarter in recurring revenue, that’s a ~500m run rate, and NTNX’s mkt cap to recurring revenue ratio is ~15.

Not trying to be trite here but it’s worth looking at. Especially since the Non-portable Software was only up 16% YoY.

Bear

2 Likes

Brandon,
Over 100 recommendations says it all
Thank you for the write-up. I finally get it
Erik

Like many, I got a lot out of Brandon’s write-up and therefore gained confidence to restart my NTNX position earlier this week. Many props to Brandon re: that piece of work by him.

Two beginner’s mind questions:

(1) does Nutanix software sit as an application on top of the Operating System (Windows Server, Linux, etc.,)? Or is it a replacement Operating System? Suspect the former ---- yet looking for that validation. And if it is the former, I totally get the size of the opportunity ---- and why the Morgan Stanley analyst (vis-a-vis Bert) might have said that Nutanix was the most innovative company she had seen in ~20 yrs of experience.

(2) Microsoft Windows Server OS does well in the data center (AWS, Azure, or on premise). Any idea if their own HCI solution is of concern? https://www.microsoft.com/en-us/cloud-platform/software-defi…

does Nutanix software sit as an application on top of the Operating System?

Yes indeed. Nutanix is not a fully functioning operating system that would run an Apple or Dell computer on its own. Unfortunately I cannot answer your second question as I am not familiar with what Microsoft truly offers in this area.

Thank you to everyone who responded kindly to my original post. I greatly enjoyed reading the Nutanix bible, and trying to make sense of what the company truly did.

Brandon

2 Likes

Perhaps one of our Nutanix experts can address this question.

My background for some years has been in high transaction volume ERP type applications. Everything I have seen from all of the DBA types I know is that high transaction volume DB applications do not do well in virtual environments or with storage other than on the same machine as the processor(s). Low to medium transaction volume applications can do OK on these environments, although the issue can arise, spectacularly, when one has to do something like a restore from backup due to some fault, so that what was a low transaction volume application is suddenly trying to do monstrous transaction volume in the restore.

Does Nutanix provide a recognition of these issues such that one can have a unified management environment which includes dedicated machines with local storage when that is the right hardware for the specific application requirements.

2 Likes

Excuse the duplicate post, I meant to post the following here rather than the other, similar thread.

I’m new on this board, so I may have missed some earlier discussions.

Here’s what I think is a good link on HCI https://www.zdnet.com/article/what-hyperconvergence-is-how-i…

The end goal of HCI is the SDDC, which consists of three legs: virtual compute, virtual storage and virtual networking. Virtual compute is established technology. So is virtual storage, although a little less so. There’s a lot of competition in both as it is being commodotized. As far as I know, only VMware has virtual networking pat down with NSX and the likes of Cisco, Microsoft and a few smaller companies playing catch-up.

With the one important leg missing, how is Nutanix going to stay competitive long term? Especially when competing in the hybrid-cloud and cross-cloud space.

Mark

1 Like

Explaining a complicated technical issue so that laymen can understand it is quite an achievement. Well done!

On an even more basic level, in the digital world anything that can be done in hardware can be done in software and anything that can be done in software can be done in hardware. Since day ONE there has been an ongoing battle between the two. EMC took the storage business away from IBM by using sophisticated software on cheap disks.

Nutanix seems to have lots of competitors

Nutanix Competitors and Alternatives
Top products also considered by reviewers before purchasing Nutanix

https://www.gartner.com/reviews/market/hyperconverged-infras…

While there should be high switching costs I don’t see much network effect, everyone can run their own brand of HCI. What makes Nutanix number ONE?

Denny Schlesinger

5 Likes

Denny:
Explaining a complicated technical issue so that laymen can understand it is quite an achievement. Well done!

Concur. Then:

What makes Nutanix number ONE?

Denny Schlesinger

IPO’ing a little over two years ago, the company is actually coming up on ten years of age. It has not shown a profit in recent reports. What is the forecast for when they WILL be in the black?

Ten years is not a trivial time span in the S/W tech jungle, especially if they’re to be considered “Number One”.