What’s .NEXT 2017 – AHV Turbo Mode

Back in 2015 I wrote a series titled “Why Nutanix Acropolis Hypervisor (AHV) is the next generation hypervisor” which covered off many reasons why AHV was and would become a force to be reckoned with.

In short, AHV is the only purpose built hypervisor for hyper-converged infrastructure (HCI) and it has continued to evolve in terms of functionality and maturity while becoming a popular choice for customers.

How popular you ask? Nutanix officially reported 23% adoption as a percentage of nodes sold in our recent third quarter fiscal year 2017 financial highlights.

Over the last couple of years I have personally worked with numerous customers who have adopted AHV especially when it comes to business critical applications such as MS SQL, MS Exchange.

One such example is Shinsegae who is a major retailer running 50,000 MS Exchange mailboxes on Nutanix using AHV as the hypervisor. Shinsegae also runs MS SQL workloads on the same platform which has now become the standard platform for all workloads.

This is just one example of AHV proven in the field and at scale to have the functionality, resiliency and performance to support business critical workloads.

But at Nutanix we’re always striving to deliver more value to our customers, and one area where there is a lot of confusion and misinformation is around the efficiency of the storage I/O path for Nutanix.

The Nutanix Controller VM (CVM) runs on top of multiple hypervisors and delivers excellent performance, but there is always room for improvement. With our extensive experience with in-kernel and virtual machine based storage solutions, we quickly learned that the biggest bottleneck is the hypervisor itself.


With technology such as NVMe becoming mainstream and 3D XPoint not far behind, we looked for a way to give customers the best value from these premium storage technologies.

That’s where AHV Turbo mode comes into play.


AHV Turbo mode is a highly optimised I/O path (shortened and widened) between the User VM (UVM) and Nutanix stargate (I/O engine).

These optimisation have been achieved by moving the I/O path in-kernel.












Just kidding! In-kernel being better for performance is just a myth, Nutanix has achieved major performance improvements by doing the heavy lifting of the I/O data path in User Space, which is the opposite of the much hyped “In-kernel”.

The below diagram show the UVM’s I/O path now goes via Frodo (a.k.a Turbo Mode) which runs in User Space (not In-kernel) and onto stargate within the Controller VM).


Another benefit of AHV and Turbo mode is that it eliminates the requirement for administrators to configure multiple PVSCSI adapters and spread virtual disks across those controllers. When adding virtual disks to an AHV virtual machine, disks automatically benefit from Nutanix SCSI and block multi-queue ensuring enhanced I/O performance for both reads and writes.

The multi-queue I/O flow is handled by multiple frodo threads (Turbo mode) threads and passed onto stargate.


As the above diagram shows, Nutanix with Turbo mode eliminates the bottlenecks associated with legacy hypervisors, one such example is VMFS datastores which required VAAI Atomic Test and Set (ATS) to minimise the impact of locking when the numbers of VMs per datastore increased (e.g. >25). With AHV and Turbo mode, every vdisk has always had it’s own queue (not one per datastore or container) but frodo enhances this by adding a per-vcpu queue at the virtual controller level.

How much performance improvement you ask? Well I ran a quick test which showed amazing performance improvements even on a more than four year old IVB NX3450 which only has 2 x SATA SSDs per node and with the memory read cache disabled (i.e.: No reads from RAM).

A quick summary of the findings were:

  1. 25% lower CPU usage for the similar sequential write performance (2929MBps vs 2964MBps)
  2. 27.5% higher sequential read performance (9512MBps vs 7207MBps)
  3. A 62.52% increase in random read IOPS (510121 vs 261265)
  4. A 33.75% increase in random write IOPS (336326 vs 239193)

So with Turbo Mode, Nutanix is using less CPU and RAM to drive higher IOPS & throughput and doing so in user space.

Intel published “Code Sample: Hello World with Storage Performance Development Kit and NVMe Driver” which states “When comparing the SPDK userspace NVMe driver to an approach using the Linux Kernel, the overhead latency is up to 10x lower”.

This is just one of many examples which shows userspace is clearly not the bottleneck that some people/vendors have tried to claim with the “in-kernel” is faster nonsense I have previously written about.

With Turbo mode, AHV is the highest performance (throughput / IOPS) and lowest latency hypervisor supported by Nutanix!

But wait there’s more! Not only is AHV now the highest performing hypervisor, it’s also used by our largest customer who has more than 1750 nodes running 100% AHV!


Dare2Compare Part 5 : Nutanix can’t claim single screen management w/o extra fees or GUIs

If you’ve not read Parts 1,2,3 and 4, we have already proven several claims by HPE Simplivity regarding Nutanix to be false, as well as explored the misleading way in which HPE SVT promote data efficiency.

The fun continues and in Part 5 we will discuss HPE’s claim that Nutanix does not have a “single screen management” (by which I assume they mean Single Pane of Glass) without extra fees or GUIs.

Unfortunately the URL was not working in the HPE tweet, I responded and made HPE aware of this so I could review specifically what they are claiming, but the link at the time of writing is still not working.

It’s funny HPE SVT mention this because Nutanix is the only HCI product which has a built in, distributed, scalable and multi hypervisor management solution.

The fact Nutanix has its own interface is a huge advantage especially because Nutanix is not dependant on any 3rd parties (e.g.: VMware vCenter) to install/configure and manage our platform. This reduces cost,complexity,risk,operational tasks and the list goes on.

Nutanix “PRISM Element” HTML 5 GUI is built into every Nutanix solution regardless of hypervisor or underlying hardware. The below screenshot shows the built in management capabilities to upgrade the Nutanix Acropolis (AOS) storage layer, the built in, scale out file server, the hypervisor (ESXi, Hyper-V or AHV) as well as upgrade Firmware, our Container support and our built in cluster imaging tool, Foundation.


This means regardless of hypervisor, many of the critical tasks can be performed straight within PRISM and does not require the long in the tooth VMware Update Manager (VUM) which is long overdue for an overhaul. In fact, Nutanix supports four (4) hypervisors using our management tool (PRISM) whereas HPE SVT only has GA support for ESXi.

For customers using Acropolis Hypervisor (AHV), 100% of the management can be performed within PRISM Element and central management of multiple clusters is performed through PRISM Central.

AHV comes with all Nutanix solutions at no extra cost regardless of hardware choice (including HPE Proliant). This means customers enjoy the benefits of the next generation hypervisor, designed and built for HCI and Enterprise Cloud.

Unlike HPE SVT for example, Nutanix does not have a limit of 8 nodes per datacenter or 32 per “federation”, PRISM element can support a cluster of any size (currently no support limits) and PRISM central manages all the clusters.

Nutanix management is not tied to or more importantly dependant on VMware vCenter or any other hypervisor management tool, which adds to the resiliency and simplicity of the Nutanix platform. PRISM automatically scales in both performance and resiliency as a cluster expands to ensure consistent performance for system administrators. This avoids the complexity of designing/installing and maintaining a highly available vCenter solution which also uses additional compute and storage resources.


  1. Nutanix PRISM Element GUI is built in and comes included with every Nutanix deployment
  2. Nutanix PRISM is not limited by the number of nodes it can manage
  3. PRISM Central is used to manage multiple Nutanix clusters centrally if required but is not mandatory.
  4. Nutanix provides at no cost the next generation hypervisor (AHV) which has 100% of all management performed within PRISM GUIs.
  5. AHV eliminates the requirement for Hypervisor licensing (e.g.: VMware vSphere) which actually reduces overall costs, this is unique to Nutanix.
  6. PRISM supports 4 hypervisors (ESXi , Hyper-V, AHV and XenServer) which delivers a consistent management interface for multi-hypervisor environments which are becoming more and more common.

Many of the above points are unique to Nutanix and have been designed and built to be a truly webscale platform, not a ROBO/SMB or <32 node solution. Nutanix can start small and continue to scale to any size, with the PRISM Element management stack automatically scaling to suit as nodes are added.

Return to the Dare2Compare Index:

Problem: ROBO/Dark Site Management, Solution: XCP + AHV


Remote office / Branch Office commonly referred to as “ROBO” and dark sites (i.e.: offices without local support staff and/or network connectivity to a central datacenter) are notoriously difficult to design, deploy and manage.

Why have infrastructure at ROBO?

The reason customers have infrastructure at ROBO and/or Dark Sites is because these sites require services which cannot be provided centrally due to any number of constraints such as WAN bandwidth/latency/availability or, more frequently, security constraints.


Infrastructure at ROBO and/or dark sites need to be functional, highly available and performant without complexity. The problem is as the functional requirements of the ROBO/dark Sites are typically not dissimilar to the infrastructure in the datacenter/s, the complexity of these sites can be equal to the primary datacenter if not greater due to the reduced budgets for ROBOs.

This means in many cases the same management stack needs to be designed on a smaller scale, deployed and somehow managed at these remote/secure sites with minimal to no I.T presence onsite.

Alternatively, Management may be ran centrally but this can have its own challenges especially when WAN links are high latency/low bandwidth or unreliable/offline.

Typical ROBO deployment requirements.

Typical requirements are in many cases not dis-similar to those of the SMB or enterprise and include things like High Availability (HA) for VMs, so a minimum of 2 nodes and some shared storage. Customers also want to ensure ROBO sites can be centrally managed without deployment of complex tooling at each site.

ROBO and Dark Sites are also typically deployed because in the event of WAN connectivity loss, it is critical for the site to continue to function. As a result, it is also critical for the infrastructure to gracefully handle failures.

So let’s summarise typical ROBO requirements:

  • VM High Availability
  • Shared Storage
  • Be fully functional when WAN/MAN is down
  • Low/no touch from I.T
  • Backup/Recovery
  • Disaster Recovery


Nutanix Xtreme Computing Platform (XCP) including PRISM and Acropolis Hypervisor (AHV).

Now let’s dive into with XCP + PRISM + AHV is a great solution for ROBO.

A) Native Cross Hypervisor & Cloud Backup/Recovery & DR

Backup/Recovery and DR are not easy things to achieve or manage for ROBO deployments. Luckily these capabilities are built-in to Nutanix XCP. This includes the ability to take point in time application consistent snapshots and replicate those to local/remote XCP clusters & Cloud Providers (AWS/Azure). These snapshots can be considered backups once replicated to a 2nd location (ideally offsite) as well as be kept locally on primary storage for fast recovery.

ROBO VMs replicated to remote/central XCP deployments can be restored onto either ESXi or Hyper-V via the App Mobility Fabric (AMF) so running AHV at the ROBO has no impact on the ability to recover centrally if required.

This is just another way Nutanix is ensuring customer choice and proves the hypervisor is well and truely a commodity.

In addition XCP supports integration with the market leader in data protection, Commvault.

B) Built in Highly Available, Distributed Management and Monitoring

When running AHV, all XCP, PRISM and AHV management, monitoring and even the HTML 5 GUI are built in. The management stack requires no design, sizing, installation , scaling or 3rd party backend database products such as SQL/Oracle.

For those of you familiar with the VMware stack, XCP + AHV provides capabilities provided by vCenter, vCenter Heartbeat, vRealize Operations Manager, Web Client, vSphere Data Protection, vSphere Replication. And it does this in a highly available and distributed manner.

This means, in the event of a node failure, the management layer does not go down. If the Acropolis Master node goes down, the Master roles are simply taken over by an Acropolis Slave within the cluster.

As a result, the ROBO deployment management layer is self healing which dramatically reduces the complexity and and all but removes the requirement for onsite attendance by I.T.

C) Scalability and Flexibility

XCP with AHV ensures than even when ROBO deployments need to scale to meet compute or storage requirements, the platform does not need to be re-architected, engineered or optimised.

Adding a node is as simple as plugging it in, turning it on and the cluster can be expanded not disruptively via PRISM (locally or remotely) in just a few clicks.

When the equipment becomes end of life, XCP also allows nodes to be non-disruptively removed from clusters and new nodes added, which means after the initial deployment, ongoing hardware replacements can be done without major redesign/reconfiguration of the environment.

In fact, deployment of new nodes can be done by people onsite with minimal I.T knowledge and experience.

D) Built-in One Click Maintenance, Upgrades for the entire stack.

XCP supports one-click, non-disruptive upgrades of:

  • Acropolis Base Software (NDSF layer),
  • Hypervisor (agnostic)
  • Firmware
  • BIOS

This means there is no need for onsite I.T staff to perform these upgrades and XCP eliminates potential human error by fully automating the process. All upgrades are performed one node at a time and only started if the cluster is in a resilient state to ensure maximum uptime. Once one node is upgraded, it is validated as being successful (Similar to a Power on self test or POST) before the next node proceeds. In the event an upgrade fails, the cluster will remain online as I have described in this post.

These upgrades can also be done centrally via PRISM Central.

E) Full Self Healing Capabilities

As I have already touched on, XCP + AHV is a fully self healing platform. From the Storage (NDSF) layer to the virtualization layer (AHV) through to management (PRISM) the platform can fully self heal without any intevenston from I.T admins.

With Nutanix XCP you do not need expensive hardware support contracts or to worry about potential subsequent failures, because the system self heals and does not depend on hardware replacement as I have described in hardware support contracts & why 24×7 4 hour onsite should no longer be required.

Anyone who has ever managed a multi-site environment knows how much effort hardware replacement is, as well as the fact that replacements must be done in a timely manner which can delay other critical work. This is why Nutanix XCP is designed to be distributed and self healing as we want to reduce the workload for sysadmins.

F) Ease of Deployment

All of the above features and functionality can be quickly/easily deployed from out of the box to fully operational ready to run VMs in just minutes.

The Management/Monitoring solutions do not require detailed design (sizing/configuration) as they are all built in and they scale as nodes are added.

G) Reduced Total Cost of Ownership (TCO)

When it comes down to it, ROBO deployments can be critical to the success of a company and trying to do things “cheaper” rarely ends up actually being cheaper. Nutanix XCP may not be the cheapest (CAPEX) but we will be the lowest TCO which is after all what matters.

If you’re a sysadmin and you don’t think you can get any more efficient after reading the above than what you’re doing today, its because you already run XCP + AHV :)

In all seriousness, sysadmin’s should be innovating and providing value back to the business. If they are instead spending any significant time “keeping the lights on” for ROBO deployments then their valuable time is not being well utilised.


Nutanix XCP + AHV provides all the capabilities required for typical ROBO deployments while reducing the initial implementation and ongoing operational cost/complexity.

With Acropolis Operating System 4.6 and the cross hypervisor backup/recovery/DR capabilities thanks to the App Mobility Fabric (AMF), there is no need to be concerned about the underlying hypervisor as it has become a commodity.

AHV performance and availability is on par if not better than other hypervisors on the market as is clear from several points we have discussed.

Related Articles:

  1. Why Nutanix Acropolis hypervisor (AHV) is the next generation hypervisor
  2. Hardware support contracts & why 24×7 4 hour onsite should no longer be required.