topic Re: NFVIS for UC in Unified Communications Infrastructure

NFVIS for UC

JamesHawkins — Mon, 23 Feb 2026 10:38:26 GMT

Hi,

I am interested in the NFVIS for UC hypervisor that was announced last years.

Does anyone know when this is actually going to be available? I saw a document that said support for UC would be in version 4.18.1 but I cannot see anything about such support in the Readme file on the download page.

Also does anyone know how the commercial side works? Is NFVIS for UC something that needs to be purchased? If so does that also apply to customers with BE6000/BE7000 bundles which were supplied with perpetual licenses for VMware ESXi?
The lack of information about this is frustrating as lots of customers are concerned about vulnerabilities in ESXi which cannot be patched without throwing $$$$$ at Broadcom.

Re: NFVIS for UC

Jason Bandy — Tue, 24 Feb 2026 17:43:39 GMT

Eager to know this info as well!

Re: NFVIS for UC

emiliolara — Thu, 26 Feb 2026 02:44:38 GMT

"Cisco-NFVIS-for-UC is NOT yet orderable or TAC-supported and will be made available shortly" - Published Date: 2026-02-05

https://www.cisco.com/c/dam/en/us/td/docs/voice_ip_comm/uc_system/Solution-guide-virtualization-guide/Cisco-Virtualization-Guide-for-Cisco-On-premises-Calling-Applications.pdf

"Cisco NFVIS-for-UC is targeted for beta in Q4 CY2025 and general availability in Q1 CY2026" - On Oct 29, 2025

https://blog.webex.com/collaboration/expanding-flexibility-for-cisco-ucm-with-nfvis-for-uc/

So, probably in march 2026.

Re: NFVIS for UC

JamesHawkins — Thu, 26 Feb 2026 11:45:59 GMT

Thanks there is some really useful technical information in those links.
It is still frustrating that the commercial information is not yet available. Let's hope it is released by the end of Q1.

Re: NFVIS for UC

shannong — Tue, 17 Mar 2026 12:35:13 GMT

NFVIS-for-UC is available in CCW now. A-NFVIS-COLLAB

Re: NFVIS for UC

JamesHawkins — Tue, 17 Mar 2026 12:50:27 GMT

Thank you, CCW says it is $420 list per month or $5,040 per year!
I was hoping and expecting that it would be free for customers who have BE6000 or BE7000 support contracts as those used to include the ESXi hypervisor.
Not a great message for those customers who want to stay with Cisco on-prem.

Re: NFVIS for UC

shannong — Thu, 19 Mar 2026 13:48:32 GMT

Since M6, customers always needed to buy ESXi and support for it on their own. Of course, now ESXi is subscription based only so there is no separate support to purchase.

Re: NFVIS for UC

JamesHawkins — Tue, 17 Mar 2026 14:08:22 GMT

When the BE6000 and BE7000 launched and for several years afterwards they included perpetual licenses for VMware ESXi.
Some versions were even tweaked to provide API hooks to allow Prime Collaboration Deployment to automate installs, upgrades etc.

After Broadcom acquired VMware whatever deal Cisco had with them was ripped up and and it became a necessity to be fleeced by Broadcom for VMware subscriptions. Cisco also moved to the Flex plan subscriptions rather than perpetual licenses at around the same time.

Re: NFVIS for UC

johnirey1 — Wed, 18 Mar 2026 18:14:07 GMT

Extremely disappointed in this price point. The execution of this rollout has been a disaster. Customers aren't going otherwise look at this and say, "I guess I will just do Webex Calling". They are going to look hard at other vendors.

Re: NFVIS for UC

jarias — Wed, 18 Mar 2026 19:54:08 GMT

Batch reply to the thread.

Cisco on-premises calling support of Cisco NFVIS-for-UC and Cisco Compute Hyperconverged with Nutanix recently went FCS.

Support has been rolling out in phases. If you've been tracking the customer/partner/user group webinars and the recently posted tech docs, then you know what is FCS right now:

Applications
- 15 SU4a of UCM, SME, Enhanced Survivability Node for Webex Calling dedicated instances
- 15 SU4 of IMP, CER (Unity Connection planned, still WIP, see the UCM SU4 updates for target FCS)
- X15.4 of Expressway
Hypervisors/Hardware
- Cisco NFVIS-for-UC 4.18.2a on calling appliances (BE6K/7K M5/6/7 and CE1400V M7). Commercial offer A-NFVIS-COLLAB, do NOT use any others like L-LIC-NFVIS or (EOL announced) FL-NFVIS-*.
- Nutanix AHV 10.0 / AOS 7.0 on Cisco UCSC/HCI-Series, UCSX/HCIX-Series, UCS B-Series, HX-Series supported for Cisco Compute Hyperconverged with Nutanix, either as Cisco HCI Node or Cisco Compute-only Node integrated with Cisco HCI Node.

The code on NFVIS 4.18.2a was ready earlier than the commercial offer and that was understandably frustrating...for us as well. We decided to still publish the technical information so partners could immediately start working designs and migration SOWs due to many customers with urgent migrations in the spring. A-NFVIS-COLLAB is now orderable; as indicated in last year's webinars to user group, customers and partners, paid subscription (can't be free as there are lots of new costs to offset plus we need to ensure we can sustain this for as long as UCM will be around). Offer pricing rationalizes vs. what BE6K/7K customers paid pre-Broadcom and what NFVIS can (not) do vs. other hypervisors. We are seeing excellent response on this from the target market segment.

Some other history ... the BE6K/7K-specific embedded OEM offers of VMware ESXi went away pre-Broadcom with the M6 appliance generation. That was done due to changing customer requirements where they outgrew those legacy offers and needed regular (pre-Broadcom) vSphere Standard Edition or Enterprise Plus Edition features. Back then there was capability to reference-sell Cisco UCS pricelist OEM offers until post-Broadcom, when they changed channel partner structure including cancellation of all Cisco OEMs. Cisco NFVIS-for-UC is now what you should look at for BE6K/BE7K appliances, as Cisco can no longer sell/renew any VMware products.

It's certainly an exceedingly complex business situation. "All apps run on ESXi which runs on all hardware" was a de facto industry standard for 15 years, and taken for granted like breathable air as in majority of customers. Now it's off-limits for many customers , and they are dealing with messy middle of what to migrate too, which may require revisiting partly cloudy, multi-hypervisor, multi-hardware platform if they don't accept what VMware looks like today.

Hope that helps!

-james

Re: NFVIS for UC

shannong — Thu, 19 Mar 2026 14:33:56 GMT

Or instead, Cisco could've worked on supporting Hyper-V. It only runs ALL of Azure and O365. Or the DECISION (no technical work involved) could have been made to support Nutanix on not just its own servers. So silly and frustrating.

Re: NFVIS for UC

Al Reid — Thu, 19 Mar 2026 15:13:23 GMT

Reading comment and feedback have a lot of opinion about this and understand Broadcom with the VMware licensing model started all of this madness. Cisco have offered alternatives but still kept this very proprietary and Cisco centric for business reason I totally understand and will likely generate a bunch of new revenue as a result.

The dream scenario would have been Cisco support any Linux KVM based hypervisor but that's a pipe dream Cisco would never allow or support.

Very frustrating few years with this and have to say I'm not loving what I've seen of NFVIS-for-UC as the interface is clunky and network interfaces and virtual networks as super confusing.

Re: NFVIS for UC

JamesHawkins — Thu, 19 Mar 2026 16:28:11 GMT

James,

Thank you for taking the time to post such as comprehensive answer - it is most appreciated.

I have not been great at attending Cisco webinars recently so may have missed updates which would have answered my query.
Information was not easy to find by searching the Cisco web site which was a bit frustrating especially when we have customer infosec teams shouting about vulnerabilities in ESXi 7.0 and their server teams pushing back on the cost of new VMware subscriptions.
Is it possible to order the NFVIS-for-UC license as an option under a Flex Plan in the same way that things like CUBE licenses can be ordered?
If not it might be worth considering as it would make the conversation with customers easier as everything thens falls under one subscription.
Thanks again for your response.

Re: NFVIS for UC

emiliolara — Thu, 19 Mar 2026 17:28:46 GMT

Whats the name of the NFVIS-for-UC to download on cisco.com?

Re: NFVIS for UC

jarias — Thu, 19 Mar 2026 17:46:58 GMT

Attempting another batch reply...

Customer attitude towards VMware by Broadcom varies globally. There are customers wanting to stay and leave. Broadcom has made a lot of changes to pricing, commercials, channels structure, architecture, etc. Some of those changes were abrupt or unanticipated. There are customers ok with those changes and customers not ok with those changes.

For the folks seeking alternative solutions ... like it or not, the computing landscape remains incredibly fragmented across commercial/open-source hypervisor/container solutions and public cloud infrastructure stacks. So if set aside VMware (which was 70% market share / 80% customer acceptance), we still lack 1-2 stacks that 80% of customers will accept. All the tech stacks are different, meaning apps won't work unmodified from one to the other, so unrealistic to try and enable/sustain/support apps across 6-12 or more different solutions to get 80% coverage.

Hypervisor strategy for onprem collaboration was deliberately biased to KVM-based solutions as that best positioned for multiple stacks. Until this part of the industry gets more standards-based, or a new normalization solution emerges, there will be challenges with trying to get "all OSes/all HW on one hypervisor". We've had this problem before ... folks may remember the problem in the 1990s was how to run apps unmodified on different compute/storage vendor technologies and stacks ... that wasn't possible (despite the HAL of every OS back then, still lots of BIOS/FW/driver issues and protocol issues and etc.) until VMware and a LOT of ecosystem co-engineering solved. Who's going to solve the problem of normalizing all the hypervisors and public cloud stacks?

On Cisco NFVIS-for-UC...

That is only trying to be a limited-purpose solution. Recall it came from the campus networking business and was built for admin by "Network Ned/Nancy" persona. It is NOT equipped, positioned as or trying to be a general-purpose replacement for VMware (general-purpose means support all apps/OSes/hardware in the universe, with advanced features like centralized management, live VM migration, etc.). It is deliberately only solving for UC-centric, calling appliance-friendly deployments admin'd by "Telecom Tom/Tina" persona. Great fit for LOTS of customers, but not all. That's why we also have Nutanix.
Commercially, A-NFVIS-COLLAB needs to be its own toplevel. We did explore "should we stuff it into Collaboration Flex and/or Business Edition 6000/7000 toplevels" but business decision not to do those, "because reasons" I won't bore with here. While there are also cross-technology commercial offers like Cisco EA 3.0, there aren't any plans currently to include NFVIS-for-UC in those.
I agree the Cisco technical docs for NFVIS could be clearer vis-a-vis what folks are used to on the UCM side. There's an active conversation on this for 26.2 release. If you have strong opinions here, relay to your partner or account team and they can get to the folks working on that.

On Nutanix hardware support for Collaboration apps, if anyone has seen the webinars...

The vision/long-term goal is land on a Nutanix HCL-based approach like we had with VMware HCL-based approach. This is how we get 3rdparty hardware coverage, as zero plan to go there with Cisco NFVIS-for-UC. We of course need a way to run on 3rdparty hardware ... for the same business reasons as when we went baremetal to VMware.
But crawl, walk, run via customer soak required due to several customer success factors that matter. Infra-layer admins frequently disagree, to which I joke "you've never met a UCM or UCCE customer before". Like it or not, calling is still crucial, and you can't cowboy it if you're in a hospital, or deployed military, or contact center line of business where $M's are lost per minute of downtime, where all IT problems manifest as end-user application symptom, with burden of proof placed on the application team with TAC SLAs for time to engage/workaround/fix. If all that didn't matter, noone would buy CUCM, they'd just deploy open-source-telephony and self-support (I'm still waiting for that to bury UCM as claimed in the 2000s).
So we have GA'd on UCS, and are awaiting customer soak. If findings are what we want, then easy to open up to other options.
We did the same crawl, walk, run with baremetal to VMware. Probably doesn't feel like it, but we are executing way faster on ESXi>NFVIS-for-UC/AHV than we ever did with baremetal>ESXi.
There are many customers understandably impatient to get the 3rdparty hardware allowed. Stay tuned.

Finally, we keep getting asks for onprem Collaboration apps to support customer-provided, open-source virtualization stacks (all kinds of stuff ... like RedHat OpenShift Virtualization, or Proxmox VE, or Harvester, or BYO distro with QEMU or Citrix/other Xen-based, or Docker/Kubernetes-based, etc. etc.). Several challenges there, even if you limit to just KVM-based solutions.

Hasn't been part of the strategy. Yes onprem collab remains, but open-source-stacks and private cloud automation stacks aren't part of the scope.
Which stack? KVM is a kernel extension, not a standard, not a product. All the Linux-based distros are slightly different, and all the KVM-based solutions have been customized. E.g. Cisco NFVIS-for-UC, Nutanix AHV, Cisco IOSXE Open Service Container, Amazon EC2's nitro, Oracle VM, OCI, Redhat OpenShift Virtualization, RHEL, SUSE, Ubuntu are all KVM-based solutions, but are pretty different when it comes to required image formats, "binary compatibility" across distros, "backwards compatibility" within a distro, guest OS support, hardware support (and who owns the HCL?), application execution/management environment & day 0/1/2 impact. Saying "two solutions are KVM-based" is like saying "Monkey and Lizard share common DNA, but don't look the same, have the same habitats, or the same diet / care & feeding, etc.".
Our customers generally require enterprise-level commercial support (even if they are not "enterprise" scale). Vs. lots of gaps here in open-source. Collab applications have a LOT of dependencies in the stack and for everything to be stable/performant/fully functional at specifically advertised capacities, all that needs to be controlled with a clear path to support with commitment to debug and fix problems. Don't believe? Reference customer reaction to our core Linux transition, or to HyperFlex or to the breakages we've had with new ESXi releases. With open-source, we don't have this ownership, and Cisco can't product-ize / appliance-ize to fix as defeats the purpose of open-source. And again unrealistic to build acres of labs to test/regress/debug everything.
Open-source is another "BYO OS" approach. In Collab enviros, customers and partners have NEVER been successful with BY O OS / BYO Server until we appliance-ized / product-ized. Happened with BYO Windows OS, BYO bare-metal servers, later with BYO ESXi and BYO UCS. Why would open-source be better/different? Vs. can't appliance-ize/product-ize like we'd normally do, as defeats the purpose of open-source. Even with a curated solution like OpenShift Virtualization, our guest OSes are not supported, so what do we do if devs need help or have a nasty TAC case? Who owns the HCL if an application symptom is isolated to hardware ?
Finally, onprem customers are getting pounded with disruption after disruption (Covid, "no longer free" Java, transcription website EOL, CentOS EOL, VMware acquisition, EKU sunset from Public CA's). More will come. Open-source doesn't sufficiently protect (good thing we did not support CentOS-based KVM as that would be dead now).
The only way we'd be able to do open-source would be via some kind of "Reference Architecture" approach, like what BroadWorks does where their "KVM" support actually means only one stack, with support demarcation application only when on that stack. Anything else customer-owned, either DIY/community support or find a commercial support vendor. As you would expect with open-source-style support models. Personally I'm not convinced customers would accept that. Like you have 10K users hard down, you call Cisco application TAC to start debugging, "you're not running on the Reference Architecture ... can you reproduce on that first?" or the usual where application is ruled out, the problem is in the KVM-based "OS" stack and/or hardware ... who do you work with to drive issue to closure?
This is why we only have commercial, closed-source solutions with enterprise-level commercial support from vendors. Historically VMware, now also Cisco NFVIS-for-UC and Cisco Compute Hyperconvergence with Nutanix.

Hope that helps!

Re: NFVIS for UC

Maren Mahoney — Thu, 19 Mar 2026 17:56:56 GMT

Another +5 to @jarias for another comprehensive answer. I'll be bringing this entire discussion back to my UC team. Thank you.

Maren

Re: NFVIS for UC

shannong — Thu, 19 Mar 2026 19:37:17 GMT

Thanks for all the added insight. Would've been nice to see a response as to why not Hyper-V. Probably about 95% of customers are running it, and it powers all of Azure/O365.

Re: NFVIS for UC

Roger Kallberg — Thu, 19 Mar 2026 19:45:24 GMT

Thank you James Arias for those comprehensive answers.

Re: NFVIS for UC

Roger Kallberg — Thu, 19 Mar 2026 21:33:34 GMT

Where is that % number coming from, is it based on your own customers or something public?

Re: NFVIS for UC

JamesHawkins — Fri, 20 Mar 2026 11:05:56 GMT

James,

Thanks again for another comprehensive reply. From my perspective the hypervisor support options Cisco have decided on make perfect sense. The majority of our customers are in the BE6000/BE7000 space so I can see that those that support larger deployments might have different opinions.

I really would not want to run Cisco UC on Hyper-V as I have been burnt by MS changes several times in the past 😭

It would be really useful if Cisco could provide partners with access to NFVIS-for-UC labs or eval licenses. Like many others I have built up a number of VMware tools and scripts that help with backups, migrations etc. and want to try to build a similar toolkit for the new hypervisor. Maybe a dCloud lab could help fill that gap.

Regarding not offering NFVIS as part of BE6000/BE7000 or a Flex Plan I am sure there were good reasons but it would have been useful when selling.
If bundled with the BE6000/BE7000 it would have been possible to set the cost of the hypervisor according to the server size. At the moment it seems that the cost to install on a 12 core BE6000 is the same as for a 28 core BE7000H which seems a bit odd. I guess there is value in keeping things simple.

Thanks again for your responses.