Win with “Dell EMC VxRail & VMware VSAN Over Nutanix HCI”

Related image

Gartner has just released its first HyperConverged Infrastructure magic quadrant as of January 2018 and unsurprisingly Nutanix is leading the way. Dell EMC is lurking right at the corner just barely behind Nutanix while VMware is 3rd on the leaders board followed by HP which I wouldn’t think would have made it as leaders without Simplivity acquisition.

Having read the report thoroughly and being involved in the HCI world almost from the beginning, I find that the report holds merit and is technically genuine. The Pros and Cons raised are straight to the point and valid in my humble opinion never the less a bit lacking in terms of HCI surrounding echo system of products such as SDN, Hypervisor, Operations, Monitoring, Cloud, and DR.

Research image courtesy of Gartner, Inc.

The aim of this post is not to technically evaluate both solutions nor to compare offerings but rather to guide potential prospects such as partners and to some extent vendors on how to win with Dell EMC VxRail over Nutanix at customer HCI engagements not only from a technology perspective but also overall engagement strategy.

“If you know the enemy and know yourself, you need not fear the result of a hundred battles.” (Sun Tzu)

First and Foremost realizing that the benefit and success of your customer is the ultimate goal, it is imperative that we really believe that VxRail in VMware environments is a better fit than Nutanix which I honestly do, not only from a pure technical perspective but from an overall offering as well. An important note here is that we DON’T care how much IOPS is provided or how many services are supported or what features are added UNLESS it maps to our customer requirements and this is key. 1 Million IOPS on a single VM on AHV with 100% read is definitely cool but does it map to a customer requirement, 99.9999999% of the times NO it does not so end of story.

When starting any conversation with customers make sure to state that many vendors have many differentiating features which is a fact never the less what really matters is what is relevant to your environment, requirements, and needs.

There is no denying the level of politics involved when approaching Nutanix and VMware relationship though many choose to deny it and beat around the bush. Dell EMC VxRail offering is based on VMware VSAN which is from an HCI Software Defined Storage perspective a rival to Nutanix AOS, both of which are the brains of the whole operation. The reason we compare VxRail with Nutanix offering not VSAN, is that VSAN alone or VSAN Ready Nodes as well do not provide the end to end SDS solution customers look for in HCI unified offerings such as Automation/Orchestration/Support/Self Service ex: VxRail Manager and Prism among others.

Needless to say that Nutanix consumerized the HCI SDS market and are clear leaders in that area, failure to admit that is a lost battle, thus making our aim just a bit harder. Fortunately most partners driven by Nutanix based on this fact are a bit aggressive and tend to undermine competition which will always come to our advantage.  By all means this does not mean it is a lost war, on the contrary, it makes the win just that more enjoyable.

“One may know how to conquer without being able to do it. ” (Sun Tzu)

I will list major points projected by Nutanix to partners and customers alike in an effort to rebuttal widely used arguments while detailing much needed customer facing skills to expose them

“If your opponent is of choleric temper,  seek to irritate him.  Pretend to be weak, that he may grow arrogant.” (Sun Tzu):

Listen, Listen, and Listen:

Image result for effective listening

“The art of effective listening is essential to clear communication, and clear communication is necessary to management success” (J Penney).

Listen to your customer, listen again, and keep listening, never interrupt them until they finish and I mean absolutely finish. If you begin speaking you have lost the battle, it is imperative that you understand problems, issues, and challenges faced before starting to ramble on solutions. I always assume I came right after a competing vendor (Nutanix) where the customer is already sold thus my only way in is making customers feel listened to, understood, and comfortable.

You might hear false statements about your product, exaggerated facts about your competition, and plain weird requirements, the trick is not to respond and node with a genuine smile until your customer has completely finished talking while taking comments on every point raised especially technical ones.

Expect to hear Vendor-Lockdown, v-Taxation, Data locality, hybrid Erasure Coding, File/Block Services  and a whole bunch of Nutanix specific points which we will tackle in a sec once it is clear that listening is a huge part of the overall strategy on an Nutanix convinced potential customer.

“In the midst of chaos, there is also opportunity.” (Sun Tzu)

Acknowledge, Acknowledge, and Acknowledge:

Image result for acknowledge

“It is the peculiar and perpetual error of the human understanding to be more moved and excited by affirmatives than negatives.” (Francis Bacon)

Listening part is over and its time to talk right, pause for a second to read your notes, and start re-affirming whatever has been said, like it or not it is just how the human mind works. For some reason 99% of us tend to believe, hold on, and defend whatever we heard first at least to some extent and we never want to proved wrong in a direct way.

Affirm and Re-Affirm whatever has been said even more agree and provide points to prove the customer perception on whatever has been discussed. It is this disruption that is going to gain the attention and trust of your customer thus allowing you to be credible in whatever is to be discussed later on.

“Engage people with what they expect; it is what they are able to discern and confirms their projections. It settles them into predictable patterns of response, occupying their minds while you wait for the extraordinary moment — that which they cannot anticipate.” (Sun Tzu)

But, But, and But:

Image result for but

“Everything before the word “but” is horse shit.” (John Snow)

After trust has been gained and genuinity has been established, it is time for the “but” statement BUT be very careful in what comes out of your mouth after the word BUT. It should not contradict with statements acknowledged or re-affirmed, the whole purpose of the word but here is to incite suspicion and stimulate curiosity.

Fortunately Nutanix partners are very aggressive and mostly not aware of their competition except what they read in Nutanix provided battle cards which to us is of great advantage. Outdated and misunderstood information relayed to customers especially VMware customers (our only targets) on VxRail and VSAN from Nutanix partners is always a welcome opportunity to discredit and prove wrong, this always leads to mistrust on behalf of the customer that projects to everything relayed on Nutanix technology.

“Knowing the enemy enables you to take the offensive, knowing yourself enables you to stand on the defensive.” (Sun Tzu).

Be careful when going up against Nutanix employees however because most of them are X VMware, v-Experts, or VCDX holders so they are technically very qualified never the less it is this commercial world we live in that would prevent them from saying the whole truth and trigger them into stating cliché statements about their competitors.

“When the officers are too strong and the common soldiers too weak, the result is COLLAPSE.” (Sun Tzu)

Discrediting one single point that is wrong or misleading (Effective listening is vital to catch those information which almost certainly have been said by opposing partners however small it may be) automatically puts customers in listening mode and tends to undo whatever has been said on Nutanix and VxRail from opposing parties.

Now that general approach has been laid down, remember not to fall into the same pit hall, do your work intensively, know both technologies, prepare very carefully, listen to your customer and map problems to technical solutions, pitch the story not the product, know your shit and never ever attack opposing vendors (Listen, Understand, Affirm, Acknowledge, then Rebuttal)  .

Vendor Lock-Down:

Related image

By far the most over-used term by Nutanix sales is vendor lockdown by which they mean with VxRail only VMware vSphere hypervisor is supported while Nutanix supports AHV, vSphere, Hyper-V, and XenServer.

Lets start by stating that VxRail is a joint offering between Dell EMC and VMware aimed at current or potential VMware customers as VSAN is an kernel embedded solution which obviously can only be supported on vSphere. IF customers want another hypervisor simply enough proposing VxRail is not an option rather Dell XC powered by Nutanix would be the obvious choice. I really laugh at opposing partners when this is brought up ( They tend to say it whatever the customer situation is )  because if the customer is sold on a different hypervisor, there is simply no competition to begin with as VxRail is architectured for VMware vSphere.

Now choosing a hypervisor to begin with or wanting to change a hypervisor in the future is not a decision to be taken slightly. The amount of technical effort and risk involved is huge and must be well thought through not only within the IT department but extends to different application and business teams. It is a fact that most customers use and are comfortable with vSphere thus are potential customers for VSAN and subsequently VxRail. Needless to say Nutanix first supported hypervisor was vSphere and biggest install base is vSphere … !

Case 1: Customer wants to continue with vSphere while transitioning to HCI is best candidate for VxRail. VSAN is an kernel embedded feature that is developed, supported, maintained, and updated by VMware so I am assuming they can better integrate with their own hypervisor. The only pitch here for Nutanix is price and will get back to that in a moment.

Case 2: Customer has VMware but is open to new hypervisors while transitioning to HCI. This will highly depend on the workloads running in the environment and their criticality. Many customers don’t like Hyper-V or XenServer for reasons I wont discuss here mainly too many patches and updates specifically for Microsoft Hyper-V while Nutanix Acropolis AHV until now is not officially supported by a wide range or business critical applications and is relatively a brand new hypervisor built on KVM. A simple statement like “Do you really want to run your LOB and business critical applications on a new hypervisor” is 99% of the time enough, don’t go into features not supported by AHV or are being developed, there is no need for that as of now, maybe in couple of years but for now stick to that simple statement which is by the way TRUE.

vSphere is the only guaranteed stable supported hypervisor that can run most workloads out there and everyone knows it. Based on the assessment of current workloads and requirements, it can be highlighted whether advantages of moving to a new hypervisor outweigh disadvantages considering technical risk and financial impact, most of the time the risk is not worth taking but again this all depends on customer requirements, use case, and workloads.

Case 3: Customer wants to change hypervisors and doesn’t want to be locked into vSphere alone. We have two scenarios here, if the customer wants to change hypervisor now while transitioning to HCI then simply enough VxRail is NOT an option. If the customer wants the option of changing hypervisors in the future ( Here is where Nutanix pitch excels Smile ) then yes VxRail is still a good fit.

Asking your customer when do you plan on changing your hypervisor would span up an answer directly from Nutanix representatives, “I may want to change at Anytime”. That statement is 100% false as its not realistic nor feasible in most cases. Changing an hypervisor after just purchasing and implementing a different one would require extensive planning and execution to say the least, support and compatibility of workloads alone is a nightmare let alone the risks involved, not only that but think of the technical impact of such a change from an execution perspective as well as skills matrix at your customer.

What about if customers have a single cluster with no enough space to re-image their appliance with a new hypervisor or a big cluster with RAID 5/6, should we assume that Nutanix will provide the same to accommodate the change and own the risk of doing so to production environments !? I haven’t seen environments with N + N high availability to accommodate a full production hypervisor change on the same hardware within its lifecycle.

This keeps VxRail in the game because when customers understand that hypervisor changes must be planned and executed extensively, it cannot be done on the same hardware that was purchased with a different hypervisor from the start. Again if purchasing the new hypervisor on the new hardware then VxRail is not an option, we are talking here on Hypervisor change pitch done by Nutanix to vSphere customers which is the least to say misleading and incomplete.

Nutanix Challenge: Make your point valid and tell us how many Nutanix customers changed from vSphere to another hypervisor on the same hardware within the hardware life-cycle for that company !? I doubt it to be more than 1% …

“Begin by seizing something which your opponent holds dear; then he will be amenable to your will.” (Sun Tzu)

vTaxation:

image

This is a term invented by Nutanix (Not Officially) to screw VMware by misleading customers into believing that VMware licenses are so costly that they are considered taxes rather than paying for services because alternate solutions may provide the same service with a significantly lower cost.

Lets start by the obvious shall we, Nutanix is NOT free, AHV is NOT free, and Prism is NOT free. Many don’t realize that Nutanix has a software model and license based on Starter, Pro, and Ultimate which are different in price and offer different set of features. Nutanix mask that by embedding this into their BOQ without most of the time informing customers about it so its practically invisible (Good Sales Strategy).

Nutanix does not offer independent AHV hypervisor sales thus as of now we cannot determine how much they would charge for their hypervisor if that would be the case. The cost you are paying for AHV and Prism is embedded in the Nutanix SDS license/hardware sold by them so claiming it to be free is far from the truth.

VMware started and led the hypervisor business in the same sense as Nutanix did with HCI but if we apply the same vTaxation terms on Nutanix we can claim that Nutanix is vTaxing you for software defined storage solutions while VxRail provides VSAN enterprise embedded license for free. We both know VSAN license is NOT FREE but it is embedded within the offering, same applies to Nutanix and AHV hypervisor so Nutanix stop this nonsense of a statement … Nothing is free !

“All warfare is based on deception” (Sun Tzu)

Data Locality:

A long endless debate in the HCI world exists between Distributed RAID and Data Locality architectures mainly between VMware VSAN and Nutanix. In essence each vendor twists the technology advantages/disadvantages to there own benefit (which makes sense from vendor perspective), Nutanix Data locality does not heavily depend on the network since a local copy of the data is always served from the same node a VM is hosted on until a failure occurs in which the VM started on other node reading from the network until a new local data copy is built on the new node on top of read I/O should have low latency since its not leaving the node itself while VMware VSAN distributed RAID relies heavily on the network since no locality exists except in a stretched cluster scenario and a VM is spread on different nodes reading from different disks.

Nutanix Data locality issue is mainly with DRS and to some extent maintenance mode, imagine only imagine DRS is invoked every 10 minutes and you have a cluster with 1000 VMs and 20 servers , the impact of moving say 10% of the VMs constantly using DRS fully automated on performance is huge as the data has to always follow the VM and until it has created a local copy its reading from the network.

VMware VSAN distributed RAID issue is mainly with network, VMs read from different hosts/disks at all times ( except if the data happens to be on the same node or in cache ) and all traffic is traveling through the network so that is major utilization of network resources and reliance, another issue that is being propagated in the market is that SSD speed is getting faster than 10GB networks thus very soon with NVMe the disk is going to be faster than the network so that would incur latency and reduced performance/speed.

The funny thing here is that Nutanix representatives fail to mention that a network loss obviously means inaccessibility for all your workloads regardless of the status of your SDS provided whomever that may be. Also note that 25GB networks are becoming standard now and will eventually be server standard assuming NVMe will be cheap enough to use for most customers.

Why Nutanix try to dazzle customers with Data Locality is out of my comprehension, from a customer perspective, no tangible verified  data has shown that limitation or advantages of each approach effect customer requirements and environments in 99% of the use cases. Both approaches have thousands of customers with hundreds of thousands of workloads without impact on how data is handled.

“Those skilled at making the enemy move do so by creating a situation to which he must conform; they entice him with something he is certain to take, and with lures of ostensible profit they await him in strength.” (Sun Tzu)

In-Kernel versus VSA:

In the same way Nutanix projects Data locality as a key differentiator in how data is superiorly  handled on their platform, VMware states that In-Kernel SDS functionality provided a faster IO path than Virtual Storage Appliance architecture.

The statement is 100% true not only from an IO path perspective but also from an high availability perspective ( messing with CVM would render the node useless ) . Obviously IO processed in the kernel is much faster than IO processed through a virtual machine no matter the technology used but I have to say here that no tangible verified data has shown that this impacted 99.9% of customer use cases.

Being honest and open is key for technical personnel and I find it to be in the core of real professionals. List the advantage of In-Kernel just like Nutanix pitch Data Locality but be very open about the fact that both technologies work and that IO latency introduced by VSA does not effect most workloads and is invisible to customers.

“Honest differences are often a healthy sign of progress.” (Gandhi)

Deduplication, Compression, & Erasure Coding:

All three are supported by Nutanix on All-Flash and on Hybrid appliances while on VxRail/VSAN only All-Flash appliances support Dedup, Compression, and Erasure Coding.

Though these features are supported on Nutanix never the less they do NOT come enabled by default on hybrid appliances and Nutanix does not recommend enabling deduplication on hybrid nodes or erasure coding for that matter. The overhead produced for the same does not make sense on hybrid nodes in terms of CPU, RAM , and Disk. Erasure Coding makes sense with big clusters though and its a nice feature to have available in hybrid appliances for some use cases where usable disk is more important than performance.

To my knowledge VMware is not working on supporting those features on hybrid appliances just for the fact that sales on hybrid are dropping fast and as SSD prices become cheaper, the defacto standard for HCI is All-Flash appliances. Forget about dedup/compression ratio projected by Nutanix or by VMware, it all depends on the type of data being stored.

Always pitch All-Flash appliances especially when up against Nutanix proposing hybrid appliances. Usable Disk capacity difference when using RAID1 with Nutanix and RAID5/6 with VSAN All-Flash makes All-Flash appliance sizing lower so “barely” cheap enough to compete with a hybrid Nutanix given that the right discount is offered by Dell EMC.

“These military devices, leading to victory, must not be divulged beforehand.” ( Sun Tzu )

Block & File Services:

Nutanix Acropolis File Services offers SMB CIFS highly available file shares that integrate with Active Directory and offers block services as well in terms of target ISCSI disks that can be presented to virtual machines.

VSAN 6.6 which is part of VxRail 4.5 latest release supports ISCSI target for Oracle workloads and physical hosts. It does not as of now support presenting ISCSI disks to virtual machines and Microsoft Failover in-guest clusters. VMware are working on supporting Microsoft Failover Clusters and presenting ISCSI disks to virtual machines which is supposed to be released with official support this year.

Again this depends on use cases and does not apply to all customers. If they don’t want to invest in Exchange DAG, DFS-R, and/or SQL AlwaysOn for licensing purposes , we can as of now use Server 2016 or 2012 R2 virtual machines with ISCSI target feature enabled to present shared storage to virtual machines. Host HA for that ISCSI target machine is maintained by VMware HA while Fault Tolerance (Enterprise Plus Feature) to ensure zero downtime on shared storage.

Yes This is a workaround for now and not the solution since VMware FT would still not provide service HA for ISCSI target VM. Very Very Very soon when VSAN ISCSI supports Microsoft Failover Cluster and VMs , that would be the end of block storage sharing limitations on VxRail. Since DFS-R takes double the storage, creating a guest file server cluster using ISCSI disks  presented from VSAN would solve file services as well ( The only difference here is we built the file service ( 30 min task with PS ) while Nutanix automated it with same end result.

Update: vSphere 6.7, vCenter 6.7, and vSAN 6.7 have just been released with support for Microsoft Windows Failover Clustering and Virtual Machines iSCSI shared disks which enables sharing block storage for in-guest clustering and building file services using a Windows Server File Failover Cluster. Dell EMC VxRail will support vSAN 6.7 with the release of update 1 .

Nutanix better update their Sales/Technical battle cards and Partner as soon as possible 🙂 .

Price:

Unfortunately I cannot list prices on Nutanix and VxRail for obvious reasons but I can state simply that based on my experience working with both that an VxRail All-Flash is always cheaper than a Nutanix All-Flash in term of list price even when vSphere prices are added.

Nutanix tries to play on the yearly support you have to pay VMware to present an ROI/TCO that makes sense never the less this price difference in support is most of the time an side factor knowing that most Nutanix sales are based on vSphere so that still applies …

With the right sizing and aggressive discount, almost always, VxRail will beat Nutanix when it comes down to price.

“For the wise man delights in establishing his merit, the brave man likes to show his courage in action, the covetous man is quick at seizing advantages, and the stupid man has no fear of death.” (Sun Tzu)

Conclusion:

I only touched based on major points raised by Nutanix when competing with VxRail at customer engagements. This has proven successful at multiple engagements only because being honest and sincere always pays off no matter the outcome. You might lost a deal but eventually many more would be gained because of established trust. “The art of war is of vital importance to the State.” (Sun Tzu)

“Know the enemy and know yourself; in a hundred battles you will never be in peril. When you are ignorant of the enemy, but know yourself, your chances of winning or losing are equal. If ignorant both of your enemy and yourself, you are certain in every battle to be in peril.”  (Sun Tzu)

Salam Smile .

10 thoughts

  1. “SQL AlwaysOn for licensing purposes”
    SQL BAG (Basic Availability Groups) is available in standard MS SQL licensing now FYI. No need to pay for AAG to do application-level failover.

    1. Hi John,

      BAG only supports one database per listener so impractical for enterprises where 10`s of databases exist though an option for SMBs . Also SQL 2016 and later is not an option for all customers.

    1. Hi Darren,

      Good Luck, That should depend on your requirements and no doubt about it Nutanix is a great technology.

  2. Your in kernel versus “VSA” (which no longer exists by the way) is way off. In kernel is no faster than a virtual storage controller (CVM) writing via DirectPath I/O which VMware themselves have validated a million IOPS SQL VM using this DirectPath I/O methodology. Also the entire idea of “in kernel” is invalidated when you are reading data across the network anyways as VSAN does; which is inferior to local reads as Nutanix uses. There are also many other differentiators that you miss; especially around what happens when VSAN loses a disk; versus what happens when a Nutanix cluster loses a disk. Rebuild times in VSAN are much much longer than Nutanix’s self healing capabilities and rebuilds with Nutanix happen by the ENTIRE cluster contributing to the rebuild. Too many technical differentiators to list. VSAN is very crude way to virtualize storage whereas Nutanix is an entire platform with many other services.

    1. Though this was not meant to be a pure technical comparison but I will address your statements starting by the last, you do not compare VSAN to Nutanix, you compare VSAN to Nutanix SDS services and compare VMware to Nutanix when you want to discuss other services. Would you really compare Nutanix LifeCycle Management to vRealize Automation, NSX to Nutanix SDN, Prism to vCenter with vRealize Operations, or RP4VM/SRM to AOS Replication !? I don’t think much elaboration is needed when trying to approach other services, there is no room for comparison at least at this stage, maybe in a couple of years, and do not that all are paid services and yes prices vary a lot but don’t pitch free services plz 🙂 … VSAN 6.7 data rebuilds are much faster and resilient with adaptive resync feature that ensures a fair-share of resources are available for VM I/Os and Resync I/Os during dynamic changes in load on the system, an improved vSAN storage scheduler uses congestion signaling to detect both the read and write throughput capabilities of a vSAN disk group. Also from a pure SDS services, Nutanix had an edge with AFS and I noted this but with VSAN 6.7 that has been taken care of and both shared block and file services can be natively built on VSAN though Nutanix makes it easier to do so. If you are referring to a write cache disk failure that would render the disk group in maintenance mode, well I wonder what an CVM failure would do to a node … these type of cliche statements that Nutanix have projected in the market do not work anymore, its just how VSAN works with disk groups/write cache disk and how CVM works with Nutanix, failure of any result in consequences, that is why both need N+1 at least .!? In regards to Data locality/Distributed RAID and In-Kernel/CVM , the whole point I am making is that differences in all do not effect 99% of customer user cases and both have thousands of workloads running with no performance impact so both do not have valid arguments, I don’t care if AHV can get a billion IOPS read write 64K blocks on a single VM since this will never map to a customer requirements or a realistic real world use case. I guess next your going to mention 600GB VSAN All-Flash write cache disk limitation :), let me know and we will discuss that as well when Nutanix shows us a workload on all flash that would require more than 600GB pure write caching on a 10wpd disk rather than a 1wpd knowing that VSAN hybrid doesn’t have that limitation because the write cache is used for reads and writes which would only prove that technically speaking they can support it on All Flash as well if that was a real use case.

Leave a Reply

Your email address will not be published. Required fields are marked *