Updated: Feb 8, 2020
For newbies to my blog site, this particular blog is about Hyper-Converged Infrastructure (HCI) and the true No 1 player (IMCO) in the HCI space, one Nutanix Corp currently based in San Jose, California.
The above picture is a futuristic rendering of the city of San Jose as I imagine it could be, but it won't necessarily be hovering above where it is located right now - just to give you a flavor of where I think we are going and my mindset on tech in general.
The various pages on my blog cover a wide range of geeky stuff where I dive into AI, robotics, flying, AMD processor stuff such as Threadripper/EPYC, X570/TRX40 motherboards, AMD Ryzen 3000/4000 series CPUs, SMT4, PCIe 4.0 and beyond as well as various laptop adventures I've had from time to time and some Quantum compute related shenanigans going on at UCSB and other labs I am plugged into.
I talk a lot about Networking technology, Cloudy things, SSD and NVMe Storage stuff related to server and compute systems that use PCIe 4.0 and fairly soon, PCIe 5.0.
As an ex Networking guy I have a morbid fascination for the Networking glue that makes HCI magic all come together so swell (particularly 100 GbE and beyond).
So let's start with where I think Nutanix are going in 2020.
Since I last wrote this blog page in January of 2018, a lot of things have changed for both Nutanix and myself.
Nutanix have added new dimensions to the portfolio that are very interesting and I am now at long last a Nutant myself which I have been endeavoring to become since January 5th of 2018.
That's a long emotional tale involving NetworkZilla, Nasdaq and a whole soap Opera around Springs that finally Sprung flavored HCI that left a good few people a couple of sandwiches short of a picnic that I won't get into here out of consideration for your sanity.
I'll save it for retelling if I ever get stuck in an airport some place one day and have an audience that cares who are buying me a lot of wine to hear it.
Back to Nutanix, who never take a rest from the forge of software development furnace...(like the forge of Vulcan it is in there...).
New Nutanix dimensions include various add on products and features (BEAM, CALM & FLOW) that have been added to the lineup and the existing portfolio now offers licensed capabilities for NAS (Nutanix Files), Application orchestration for cloud (CALM) and Hybrid Cloud, Object storage (Buckets) functionality and a fair few other enhancements to the current line-up such as the new Nutanix NX software for cloud!!
Kubernetes action on Nutanix through their Karbon service is gaining mind-share and Nutanix has made great strides with container tech with Karbon over the last year.
Nutanix is now a very worthy candidate for Kubernetes focused shenanigans.
The awesome Karbon Kubernetes service is an Enterprise-grade Kubernetes distribution that simplifies the operation, provisioning and lifecycle management of Kubernetes with a native Kubernetes experience.
Cyxtera is one of our early Tech Preview customers and Karbon has made a significant improvement in their business agility by being able to deploy a Kubernetes cluster in mere minutes.
I had drinks with them the other night at a spot on Santana Row and was enlightened by their awesome experiences with Nutanix Karbon. Thanks to the various folks for the update there!!
Karbon will maintain up to three different Kubernetes versions of your desire btw.
Karbon is about agility and efficiency and eliminating additional special resources to run Kubernetes.
We bring the super simplicity of the Nutanix experience to Kubernetes clusters as well with Karbon - Super semplice, molto rapido as they say in Torino.
Crystal clear clarity of operation is what Karbon delivers, so easy that you will wonder what the fuss is all about!!
On the subject of clarity around Nutanix stuff in general, I have witnessed some of our competitors and some of their NasZilla kool aid stoned channel peeps make bunk statements to their inside sales folk that Nutanix does not scale with AHV or container stuff (amongst other wild and hilarious FUD such as the Nutanix abandoning AHV tale currently circulating in the NASzilla channel).
Just so you guys know, some Nutanix customers are running 2000 nodes with AHV in a single cluster and its running just fine and we are not on AOS release 4.6 anymore folks.......
AHV meanwhile, is very definitely not being abandoned, it's shipping on over 50% of all Nutanix nodes already and can only be classed as a runaway success.
I have been a VMware bigot myself since 2004 and I can tell you that my preference if given the task to get a HCI cluster running Molto Rapido and experiencing the true One Click wonder of Nutanix would be AHV.
Where this whale of a tale FUD about AHV comes from is a thing of wonder, it really is.
I think Hypervisor envy is what makes these competitor orgs say these things because they are so tightly wound up with VMware only technology and they have no answer to our Multi-Hypervisor capability other than to make wild FUD noise about AHV.
The only problem with those competitor Hypervisors is they have a 64 node limit so for customers that want 65 or more nodes in a single cluster, they have only one option right and happily that is AHV.
Network Engineering focus for these large scale AHV deployment environments is a fact of life for anyone designing for any hyper-scale environment whether its CI or HCI.
Fortunately we have answers and awesome Networking solutions there as well and our experience with customers who have thousands of nodes in a single cluster is a fact of life chasps!! (a classic "suck it up buttercup" reality).
Other notable Nutanix developments are in the area of replication. On Premise and cloudy replication as well as major enhancements to AHV that are pretty slick.
Replication between clusters and Cloud as well as the new Nutanix NX for Public cloud (that runs in the Public cloud) is I am happy to say evolving at a rapid clip and getting more polished by the software release version via many feature advancements too long to list here.
The very cool tool to control those rabidly out of control cloud costs for public cloud offerings is in my humble opinion the best of the lot of them (BEAM) and is a very powerful tool I believe neither Nutanix folks nor our channel partners make enough use of.
The guys over at CenteredNetworks are examples of new VAR orgs that are making magnificent use of BEAM to get them into new accounts though.
BEAM is like crack for IT junkies hooked on cloud. In fact it is the common cure for Public cloud cost overruns.
When you see the mountain of $$ you can save with BEAM for your current Public cloud spend you will be totally staggered.
BEAM is the best Public cloud cost control tool I have seen to date, and I have evaluated a good few.
Even if you don't want Hybrid cloud infrastructure yet and are a public cloud only organization, you must run BEAM at the very least, it will pay for itself in 30 days flat.
Great seed action is BEAM.....Most of my customers who have been properly aligned with BEAM have also gone on to buy Nutanix clusters for DevOps and then production.
I let BEAM coolness and capability speak for itself. You will be impressed, I guarantee it!
Nutanix are now fully into the Software Defined strategy part of their game-plan and deep into the new era of many hardware platforms in the potential solution portfolio capable of running their fabulous software.
The most notable new Hardware vendor addition to the Nutanix HCL is HPe with their excellent DX and Apollo series server goodies specifically primed to run Nutanix.
Once a resistor to the Nutanix way, HPe are now fully on-board and the 19 odd HPe DX series offerings are a very welcome new addition to the Nutanix HCI cloudy things tea party through their HPe Greenlake program.
HPe Greenlake is the best thing since chariot wheels wuz invented......This is where you pay for everything OPEX style and you only pay for what you use.
This merged Nutanix Enterprise cloud with HPe DX and Apollo boxen approach is delivering on premise cloud as a service to customers with full Public cloud integration.
HPe and Nutanix partner to install everything in your data center and you get to avoid nasty career ending CAPEX and over-provisioning minefield dramas.
Great collaboration is to be seen between Nutanix and HPe and a great many others in the OEM server space as well by the way.
The HPe fare is like all the other Nutanix fare available for all leading Hypervisors (AHV, ESXi and Hyper-V) and extends to all the major cloud operators (GCP, Azure and AWS).
HPe also compete with Nutanix with two products in their stable (Simplivity and HC series stuff), but have also partnered with Nutanix because of the "More is better than less" business approach.
The fact is anyone who writes and sells x64 operating systems that run on x64 CPUs compete with each other and market success with customer thumbs ups dictates what they run on these server nodes.
Its what some call capitalism and its popular because not all OS and software systems are equal and customers want to run what works and stuff they actually love and like.
Right now we are actually back in a situation akin to the Netware days when you could run Netware on almost any server hardware that conformed to the Network card, Hard drive and memory requirements for x86 server platforms and which were on the Netware HCL.
I was a CNE and MCNE back when, specializing in IPX routing and encapsulating SNA over IPX for the ATM machines banks use, so I have fond memories of those many glorious Netware campaigns and can draw many interesting parallels between then and now.
I remember that even stuff that was not on the Novell HCL could be made to work which increased the net of Novel business capture in that era.
Many smaller Taiwanese vendors like Mustek, Mecer, ASUS and MSI et al made server systems you could integrate and sell complete with Netware fully installed and running at a significant savings to the standard server fare which resulted in margin premiums all around of the exceedingly happy variety.
While most will not get there with Nutanix in quite the same way, in theory there is little reason why you couldn't do it if you really wanted to, you just have to get the Hardware OEM you want to play with to certify with Nutanix to play that game.
We are now at the realization point where PCIe 4.0 based systems are going to make a significant impact crater for those running Nutanix nodes on PCIe 4.0 capable hardware with the AMD EPYC Rome processors in them as they are all PCIe 4.0 ready and willing.
Intel meanwhile currently have no (ZERO/NADA/ZILCH) PCIe 4.0 CPU fare and will not have any until 2021 (if they are lucky).
Mike Gutberlet, Moi and Rocco G
My chums over at the Chipzilla Optane lab have a ton of AMD Threadripper and EPYC Processors that they are using for testing of their new PCIe 4.0 Optane based SSD goodies by the way.
I took the piss out of them mercilessly for an hour or so the other day over that situation.
We are witnessing the merging of Intel Optane technology with AMD CPU platforms as we go as a result of this happy situation and I have seen quite a lot of work globally in Israel, China and Taiwan where companies with a keen interest are taking the PCIe 4.0 hardware boundaries and bottlenecks to new levels and they are using what is available to them to architect new solutions (AMD EPYC Rome 7002 series) in an eye opening manner.
For the next 2 years though, if companies want to play the PCIe 4.0 High Performance Compute game, AMD EPYC, Ryzen or Threadripper is the only game in town.
Why, it's so good that even Intel themselves use it!! Now that is what we call a classic "ZEN" moment!!
Oh the serendipity and deep Irony of it all!!
The pricing is just awesome as well by the way so it really is an all around win-win for everybody.
AMD and Intel are actually working together quite well with regards to PCIe 4.0, Optane and 3DXPoint technology based compute goodies and technology.
Right now, as far as I know, only HPe and Lenovo are doing qualifying work on their AMD EPYC platforms with Nutanix.
Dell have an AMD EPYC platform or two but are playing VxRAIL games and excluding themselves from the Nutanixverse.
Their loss. They obviously don't understand the concept, "more"....
The Wraith Queen explaining the concept of "more" to her lunch
All of this stuff gives customers choices of platforms to consider from a cost point of view when comparing maintenance and support of say Nutanix NX series HCI nodes to HPe (DX), Dell (XC) or Lenovo (HX) nodes as well as customer consideration for existing compute architecture which is probably focused around one or other of the core four server vendors already.
The Nutanix foray into offering Nutanix for HPe DL and Apollo series servers (DX) and Cisco UCS B and C series UCS platforms has been so successful that customers can now leverage their existing server vendors maintenance and support at much more competitive rates than that which Nutanix offer for their own Nutanix branded NX series platforms.
Nutanix have been moving further into the software defined world on the hardware of your choice model that allows customers to leverage their purchasing muscle with Dell, Lenovo, HPe, Cisco, Inspur, Fujitsu et al.
Today, apart from the fab five server vendors, you can in fact also run Nutanix on a great many lesser known hardware platforms such as that from Fujitsu, Inspur, Huawei and allegedly also on Hitachi Vantara hardware, so I am sure that as this all unfolds we will in fact see this happen across a broader spectrum than the current top four server vendors that they started off with.
In fact a swift glance at the latest HCL tells you the list is already pretty long.
Fujitsu, Huawei and some others I just mentioned actually have pretty big install bases in some regions of the planet and some of them are here in the US market and they too are now to be found in the Nutanix HCL.
This means that Nutanix can now cater to global regional hardware preferences and believe me they are a fact of life.
Does this mean Nutanix are getting out of the Hardware business completely?
Many of these companies who sell servers also sell HCI platforms and do everything to convince their customers that their HCI platform is better.
The real situation is some of them are such bad customer experiences that some customers have gone back to the old storage and server infrastructure model.
I would advise against giving up the Hardware development as a result so as to keep plugged in to the mainstream hardware development stuff as it becomes the latest and greatest must-have bleeding edge technology wave of the moment.
It is a great way to keep the finger on the pulse of hardware development in general, not just for servers but NVMe SSD and Networking NIC stuff as well.
There was already considerable investment in the NX series SuperMicro product line and this I suspect will continue, but over the next few years I expect this might become a legacy offering and the focus going forward will be on new generation hardware from SuperMicro, Lenovo, HPe, Dell, Cisco, Inspur and the many others who care to leap aboard the Nutanix HCI tea party bus as time goes by.
Hopefully an AMD based NX node will appear sooner rather than later.
Did you know, SuperMicro have Plenty of AMD platforms in their lineup (hint-hint)?
Historically, maintenance and support numbers have in fact been the primary reason why some Nutanix deals were lost by the way, so from a strategic point of view this expanding HCL is a welcome development.
I did have a chat with SuperMicro guys recently who told me the gap there was closing pretty dramatically so this may be yesterdays story already.
I know I am pretty pleased about that and the AMD EPYC story as it unfolds.
The problem with some OEM server vendors such as Inspur though is they are dragging their heels getting into the AMD game and it is going to really hurt them if they don't develop some acute vision pretty soon.
AMD is definitely not a temporary flash in the pan.
Get with the EPYC program guys!! I am a fan, you should be as well!!!
Only Lenovo, SuperMicro and Huawei have taken AMD seriously in the APAC region from what I see in the market space.
As the BORG queen would tell them, "Resistance is futile, you WILL be assimilated!".
We cannot take Huawei seriously here in the US market for reasons demonstrated by the current US administration, which leaves us with SuperMicro and Inspur to consider as Chinese based options (Yes I do know Taiwan is not part of China).
For years the only reason why Intel was better, despite having the worst Front Side Bus architecture in the CPU biz was their fabrication process and cache buffer trickery.
Now that AMD are the leaders in fabrication tech (7nm and soon 5nm) and they have the superior bus architecture, Intel find themselves a loooong way behind AMD on the CPU dev front.
I see the new Apple A14 CPU Architecture is 5nm so they managed to get slightly ahead of AMD there on the fabrication process front, but manufacture at scale has and is currently an issue for Apple who do not have quite the same volume demands that AMD and Intel have.
Having competition between Intel and AMD is going to be awesome for all of us though.
It's time for what they are calling the "Core Wars".
AMD leading us to 4 threads per core with SMT4 is also going to be pretty darn awesome in the forthcoming Core Wars.
Now that these pesky maintenance and support cost reasons with the big 5 in on the game going away, we can expect much more sane cost models that will benefit potential future customers, so that's a great situation for all of us to be in going forward.
Existing Nutanix customers already on SuperMicro hardware can stay on those platforms and transition to the more mainstream server vendors if they so desire or they can have clusters of HPe DX/Apollo, Dell XC, Lenovo HX and Fujitsu (FX?) with their NX clusters.
You just cannot mix hardware vendors within a cluster but there is no reason why you can't have hardware for specific use cases if you want.
For example if I wanted some of that AMD EPYC PCIe 4.0 high performance compute octane I would definitely be deploying one HPe cluster and one Lenovo cluster to see which is the best and get into the 25/100GB Ethernet network glory that comes with it while I was at it.
I would also be keen to see how Kubernetes and Karbon perform on that platform as to me EPYC Quad CPU boxen with 64 or 128 Cores featuring 4 threads per core heaven plus a fleet load of RAM strikes me as an ideal high density platform capable of dealing with Dock-loads of containers with the Nutanix simplicity as the engine core.
Now others have played the OS on Intel x86 32 bit hardware game for many years.
Microsoft for example always sold software you could load on any compatible x86 server platform from the days of 32 bit Windows NT to the present day x64 Windows Server 2019 platforms and Linux distributions have been available the exact same way.
You just had to make sure the hardware you wanted to use had the suitable OS drivers you needed and Microsoft published all the .net and API info a developer needed to write these drivers.
With Microsoft of late they have owned the drivers with the help of the hardware manufacturer and that works well for most of the hardware developers out there but there have been irksome exceptions.
In general though, from a Processor and motherboard chipset point of view you can run any OS on the hardware as long as it understood the processor and chipset instructions.
As everyone understands x86/x64 (AMD wrote the x64 code Intel uses btw), there is no problem between Intel or AMD apart from the different chipsets they each use.
That's why I laughed when I heard some Cisco reps telling customers running Nutanix AOS on their UCS servers is not supported by Cisco and a few HPe reps punting HC380 were also heard to make such silly notion in sheer desperation.
Either Cisco make x86/x64 server hardware or they don't.
That short sighted approach was a dangerous admission that their server platforms were not x64 architecture capable, which they of course are.
Nutanix AOS is a server OS, Nutanix will support it, the server vendors can hardly block an aspiring OS vendor from running their OS on their x64 server hardware now can they?
Nobody who owns and runs Microsoft Windows 2016 on a Cisco UCS server calls Cisco first when there is a Windows problem unless the Microsoft support guys tell them "yeah, you need to log an incident with Cisco because their hardware has a driver problem they need to help us fix".
Its not in Cisco's best interest to refuse to work with Microsoft as when customers hear "oh this server model can't run Windows" they will of course buy something else that can.
Since I pointed this fact of life out to a few Cisco HyperFlex die-hards that threw this dangerous logic free argument on the table, they mercifully shut that argument down rather rapidly.
What these Cisco types really mean is "hey, if you run Nutanix on our UCS servers instead of HyperFlex, we're just going to have a 2 year old super mega tantrum about it and chuck all our toys out the cot while we have a good wail about it".
Very grown up....
Nameless Cisco HyperFlex AE on hearing customer desire to run Nutanix on UCS
Other Hyper-Converged platform vendors such as Dell and Cisco are locking in their own HCI software platforms to just their hardware which I view as unfortunate and one dimensional.
This is fine for vendors who are developing their software that way, but it is not as flexible as Nutanix HCI on HPe (DX), Dell (XC) or Lenovo (HX) or any others like the Cisco UCS platforms that Nutanix is now fully behind plus any other hardware that eventually gets certified by Nutanix.
Of course you will hear Dell and Cisco sales folks claim Hyperflex and VxRAIL are just as good, if not better, but sadly this is not the case or the solution that those customers already exposed to Nutanix simplicity want.
They also do not want to be locked into just one vendor for obvious business reasons commonly labelled as "RISK".
I have run POC's for Hyperflex at select customers and the getting it Up & Running time runs to at least 5 days at best vs the 2-4 hour situation with an average 8 node Nutanix cluster and that includes all the cabling and racking time, never mind the slick experience the Nutanix interface gives you.
Simplicity so easy 8 year old kids can run it is in fact is what Nutanix is all about.
What is happening in the real world is that after POC evaluations and initial tentative small scale customer deployments, customers just go nuts for more Nutanix.
Some of my customers have reduced their OPEX costs by as much as 58% with this platform.
That's some powerful Data Center crack to get happily addicted to!
Looking at other competitors to Nutanix and swift forays into playing with these alternative platforms, you swiftly conclude that Nutanix are a mile ahead of everyone else.
I evaluated emerging technology for 5 years when I was at ePlus by the way, so I am very familiar with everything that is out and about out there.
More importantly Nutanix are not sitting on their backsides and cashing in sans keeping their fingers on the pulse of IT which is what too many tech companies have done.
Most of those are now just fond memories. Actually some are pretty gruesome nightmares as well, to be totally transparent about it.
Material for a lot of future blog postings on the doings and screwings of the IT Ewings as my grandpa used to say when I used to entertain him with the Tales of Wang, BlackBerry, 3Com, Novell, SCO and Co.
Nutanix knows full well that there is no static in IT and the only constant is change itself.
Snooze and lose. It's pretty simple.
If you move to the side and cash in without continual development and investment in R&D you are doomed to being "Yesterday's Hero".
If you need convincing of this situation may I point out one BlackBerry corp or if you are in the Bay area just go take a tour of the NASzilla campus in Sunnyvale to see how few buildings they are occupying nowadays out of that campus complex which was all NASzilla for so many years.
Google have moved in to 12 of the 16 buildings there to date.
At least Thomas and George Kurian are twin brothers so they have kept the campus in the family so to speak!!
Meanwhile, our closest competitor in the HCIverse right now, Dell-EMC are actually competing with their own Nutanix XC platform via VxRail and the VxRail 4.5 code is actually pretty good if you are a VMware only Hypervisor customer happy with limited cloud options and being locked into just Dell hardware.
This is ironic considering the Dell XC line, which is based on various Dell servers and specifically primed for Nutanix AOS are actually in the Dell-EMC HCI portfolio and which obviously can run AHV and Microsoft Hyper-V no problem because they are running Nutanix AOS.
The Dell sales folk do try awful hard to convince customers who want to run Hyper-V or AHV not to though. They sure don't get thrilled selling customers XC fare these days.
VxRail hardware could in fact run Nutanix AOS with no problem at all as well as accommodate other Hypervisors such as Hyper-V and AHV because guess what? Many of the Servers models used for VxRAIL are exactly the same as the ones used for Nutanix and if they wanted to they could replace their Hypervisor with any hypervisor..
They just do not want to do this for obvious reasons...(They own VMware and make their own servers - actually a pretty darn good reason).
I think that Dell XC is only there in the Dell lineup because Michael Dell and Dheeraj meshed very well over Nutanix and Michael was impressed......yes he was!! He told me so himself.
The things some people say in Vegas!!
Things change and sh1t happens, we get it, but still, "more" is a better business model, I think anyways!! Now "more" is not the same as "everything".
Dell knows better than most that nobody in their right mind is locking into a single vendor.
Maybe Dell should have a business unit that does what Cisco does with their CI stuff, working just with various software partners so the option for those customers to run this software on Dell hardware is an unfettered option for the folks focused on that aspect.
They can still have their core sales guys doing the Dell-EMC thang with VxRAIL but there is no reason why Dell and Nutanix cannot team to win business even if they compete against themselves with other HCI variants and other solution options available to customers.
A customer that wants to run Nutanix on Dell hardware is a customer that wants to run Nutanix on Dell hardware, no point getting all Hyperflex about it and losing ALL the business as a result of temporary insanity now is there?
I am starting to think that compromise and win-win needs to be taught as a mandatory subject in colleges and Universities.
The fact of the matter is that painting a customer wall to wall with just one vendor is a risky business for the CIO of that company and will remove the need for the dominant vendor to compete pricing wise.
We all know what happens when a vendor starts to think it's the only game in town and sky rocket their pricing to gouge out the customer eyes now don't we?
So VxRail has taken second place in the current HCI market arena, displacing SimpliVity which HPe has struggled with immensely over the last few years since they bought them.
IMHO HPe needs to throw about $3 Billion of R&D into SimpliVity, and acquire the other bits they thought they were getting but did not and hit reset.
Now that would be a worthwhile spend? I have a soft spot for them being an ex HP Labs guy myself.
I always thought the nine years of Lefthand VSA market spin would have afforded them outright purchases of a lot of worthwhile competing technology out there....
But what do I know? (my memory banks keep telling me 42...)
The Gartner MQ actually places VMware in the no 1 HCI spot which I find bizarre because their technology is just the center piece of the many HCI variants that are out there that use VMware at their HCI core with a bunch of Python or even worse, JSON scripts to create the illusion they're not vanilla vSAN.
To me a lot of this complex VMware software jambalaya is just essential vSAN posing as HCI.
A more fair accounting of whats what in the HCI zoo would involve dividing all the HCI vendors per brand name, not by what software tech they are based on?
I mean roughly 50% of Nutanix customers prefer ESXi and vCenter partnered with Nutanix AOS so its easy to add them all up that way and say VMware is the no 1 HCI vendor but its not an accurate reflection of the actual HCI product space.
That's like saying LG is the worlds no 1 electric car company just because 65% of all EV's use their very awesome LG Electric motors and superior LG battery technology.
If you look to see how many cars have an LG logo on them or try find the LG car dealership you will have a fruitless adventure designed to enhance futility itself.
With respect to other HCI competitors to take seriously HPe, like Dell had a few HCI options in their product portfolio which are still there.
The Left Hand Networks VSA that has been around for so long was, I believe a catalyst in HPe buying SimpliVity and then HPe eventually got on board with the more is better concept and joined the Nutanix HCI tea party bus.
Dell and Cisco can actually learn from HPe in this regard. There are actually folks in both companies that know and understand this game pretty well, to be fair.
Also significant for 2020 futures is the fact that the Nutanix code supports NVMe as does the new hardware from all of the top four server vendors and soon we will see AMD and PCIe 4.0 nodes hitting the HCL with HPe and Lenovo platforms leading the charge fully equipped with the new Intel PCIe 4.0 SSD goodies.
Suddenly servers with this PCIe 4.0 capability can be turned into All Flash HCI speed demons for the nodes that require that sort of insane level of performance.
Marry Intel Optane SSD with AMD EPYC and a winner is born.
This is the beauty of the Nutanix approach, you can build nodes to serve all the storage, server and performance requirements across the x64 PCIe 3.0 and 4.0 hardware spectrum starting from single nodes and stepping up to large and fast and anything in between with large numbers of them in the cluster.
This means you can effectively tailor clusters within the Data Center or Domain to do your bidding at a more precise CPU to RAM per node ratio which also actually saves you quite a bit of the good old yankee dollah.
When you run out of any of these basics, just add a node!!
Breathtaking simplicity at small bite cost factors!! What could be better than that?
I'll tell you what is better than that and it comes via huge potential cost savings with respect to OPEX.
We all know that OPEX is where the real costs are in IT.
Effectively you can run an entire Data Center based on Nutanix tech with far less resources than traditional server-storage silo's of old and focus on the applications skills and expertise instead.
That's a good place for an organization to invest its dollar in.
The other advantages with Nutanix is that by using their AHV hypervisor you can also gain significant advantages and further simplify how operations run your Data Center environment and gain impressive scale while you are at it because an AHV cluster does not have the 64 Node limits that VMware and Hyper-V clusters come with.
If you do that you also save considerable VMware virtualization dollar. No vTAX as some are calling it.
By the way AHV ain't free, its included with every purchase, use it or lose it, the choice is yours.
You can however, build clusters running Hyper-V, VMware and AHV in the same domain using namespace architectures that solve those issues in a very slick manner using DNS namespace trickery and Prism Central if you decide to approach each IT problem you are solving with purpose built Nutanix clusters for each solution set you are architecting the solution for.
For example, if you have a SAP HANA need, build an AMD EPYC based SAP HANA cluster on HPe DX/Apollo hardware and if you have Oracle and MangoDB needs, just build them their own solution cluster on say a Lenovo based AMD EPYC cluster within the data center domain, or clone yourself like clusters that can span data centers if you need to do that.
A lot of the vCenter server functionality will appear in Nutanix software updates for AHV as time goes by and it has by now mostly all been done.
There are of course also a few benefits to using AHV that are only in AHV but customers are touching the Nutanix pressure points to ensure this appears for VMware and Hyper-V as well and these features will trickle down the stack for the other Hypervisors, AHV first though.
Scale up and scale out like it was lego is the result! Build and run it with a tight well honed Nutanix skilled operations group and you are on your way to impressive OPEX savings and big warm bear hugs from the CFO.
Data Protection is also easy with both Veeam or Rubrik, maybe even Veeam and Rubrik in some cases! Veeam is a backup software solution that like Nutanix is positioned to run on mainstream hardware of your choice, virtual machine style.
I personally prefer HYCU and Cohesity. The Cohesity file system for re-hydration purposes for me is a thing of true geek wonder.
The point is that the BC/DR view is pretty darn flexible in terms of how you build it and is just what the Dr ordered for the corporate world customer base consumption model.
The world is transforming to easier to run systems featuring simplicity that is in fact very affordable, a promise worked towards for many years has actually finally arrived!
Of course, the traditional CI based folks won't just throw in the towel and wave the white flag of surrender, no sir!
What will happen is they will find ways to offer equivalent simplicity with combinations of server and storage silos specifically integrated tightly together with a GUI overlay and fancy scripts in a solution similar to that which EMC and Cisco created with VCE VBLOCK platforms, minus the insanity of pricing that VCE was famous for of course!
The software to make this all happen just needs to be written better than the CI offerings of today that can rival Nutanix software integration, features and capability.
Nutanix have also added a new cloud dimension to the portfolio.
You can now add Multi-Cloud plus their own Cloud to the vista with all of the popular private clouds on offer (GCP, AZure and AWS) and their own cloud offering called Xi LEAP.
Xi LEAP is actually much more affordable than AWS for the backup use case in the BC/DR equation and is a worthy contender for customer consideration there.
What the world wants is a Software Defined overlay that can allow the architect to put any components together in his Data Center and operate them all from one slick interface like they were all a single appliance using any hypervisor, any cloud and any hardware on the HCL!
The Openstack platforms community serving cloud applications in particular will go ga ga for this stuff!!
I think we are a very long ways off hearing the well proportioned lady sing on the subject of CI though.
Some of them new CI offerings seem to have some serious potential if developed right???
Has she sung her last song for CI? Some are saying the CI song still remains the same..
What a shame!! (not)...