Cisco Compute Hyperconverged with Nutanix – What it means for you

Compute and Virtualization

An alternate title for this blog was “Cisco divorces wife for friends-with-benefits relationship,” or some such. Unfortunately nothing I can come up with is as absurd as the spoonful of words that is Cisco Compute Hyperconverged with Nutanix. For the sake of simplicity I’m just going to refer to this solution stack as Nutanix on Cisco.

For the unaware, Cisco came out of the blue and announced they’re killing their hyperconverged (HCI) solution, HyperFlex, in favor of just reselling industry leader Nutanix.

How this affects HyperFlex customers really depends on where said customer is with adoption and product lifecycle. If you just got HyperFlex a short while ago, and are good with expected growth, there’s nothing you really need to worry about. You can continue to use HyperFlex until the end of your existing support contract which creates a scenario more akin to a traditional refresh cycle than an Earth shattering event.

The real pain point is for folks who expect, or need to, expand their HyperFlex environment. Come March 2024 that will no longer be an option. You can always just add separate compute clusters to your stack so, again, not Earth shattering.

I don’t think customers will have to worry about being outpaced by VMware upgrades either. Cisco will continue to validate any ESXi 8.0 update releases. ESXi 8.0 just went GA in October 2022. Compare that against ESXi 7.0 which went GA in April 2020 and is still being updated, or ESXi 6.5 which had about five years of updates. Sure, it’s possible Cisco could reneg on their promise or VMware could accelerate the end of 8.0 in favor of a new release. I don’t think either is likely to happen, and if anything, the Broadcom acquisition is more likely to slow down VMware’s update cycle.

Well if panicking is not a valid option, then what? How about some whiskey and thoughtful discussion on the future of HCI in your environment? Okay, okay. That was cheesy I know.

The converged vs hyperconverged discussion seems to have died down over the past few years as folks generally settled into one camp or another. Customers who’ve bought into HyperFlex, and maybe even those traditionalists with UCS, will have an opportunity to review those decisions. HCI isn’t going anywhere, but the available solution set has narrowed. The Broadcom acquisition raises questions/fear/FUD about the future of VMware which is a boon to Nutanix’ ever increasing AHV install base.

It will be interesting to see how this shift in the marketplace and vendor partnerships impact the HCI vs converged discussion over the next few years. While the general pros/cons at a technical level haven’t changed, we’ve seen (again) how a single company can, out of the blue, nuke an entire datacenter footprint and methodology.

It’s too early to say just how the Cisco and Nutanix partnership will develop. Cisco is leading this transition so heavily that all the documentation is from Cisco, and I’ve seen nary a blog article from Nutanix. It also remains to be seen how Cisco will handle the VMware question… do they continue to play well with VMware or do they start to push AHV heavily?

And no, I don’t think Cisco is going to buy Nutanix. For starters, I don’t think Cisco could really manage it with how much they’re shelling out for Splunk. More importantly this partnership really allows Cisco to have their cake and eat it too. HyperFlex, arguably, wasn’t close to the top tier HCI solution sets out there. This allows Cisco to exit that market, put their resources somewhere else, but still be able to sell a HCI solution.

For what it’s worth, I think Nutanix and AHV is great. While I’m more a proponent of traditional converged solutions there’s a place in most environments for Nutanix. I’d like to think that AHV has the potential to take an even larger chunk out of VMware’s install base, but being limited to Nutanix validated hardware is a double-edged sword. While allowing them to better optimize and deliver on the software solution, it keeps them out of areas where converged is the smarter play. Albeit for most customers the comparative factors here is like splitting hairs. Would be nice if Nutanix would at least open the door to external datastores, something I’ve commiserated with Nutanix engineers over.

One last bit for this section… If you’re considering stretching your on-prem virtualization environment to AWS or Azure, I think NC2 is a lot slicker and easier to use than VMware in AWS or Azure. Scenarios like this will also have a bearing on what decisions customers make moving forward. I have a whole long-ass blog article on this that I wrote a month ago, I just haven’t had a chance to play around with Nutanix Move to finish it off.

Nutanix on UCS

As of launch Cisco is only supporting Nutanix on C220 M6 and C240 M6 storage nodes with the right all flash configuration. New M7 nodes will be out sometime Jan/Feb 2024.

I don’t know the exact time frame, but support for UCS X-Series blades is on the roadmap for mid 2024. In my opinion that will be the ideal deployment model for Nutanix on Cisco. Whether Cisco will announce enough validated configurations to compete with the greater options Nutanix supports on the like of HEP, Dell, Fujitsu, etc. remains unseen.

At the very least I expect some graphics where the X in “UCS X-Series” is replaced with the Nutanix X logo.

Prism vs Intersight

One outstanding question is how Prism will compare to Intersight. Prism, Nutanix’s management GUI, has been around for a long while and has generally been well received. Intersight is Cisco’s management GUI and has been, well, from what I can tell mostly ignored by folks preferring to use UCSM. From what I’ve been told Nutanix on Cisco will require Intersight for management, with Prism as a secondary option. In theory, having Intersight as a full blown management plane for all things Cisco compute sounds nifty. Cisco’s ability to execute on that vision also remains unseen.

Moving from HyperFlex to Nutanix on Cisco

I want to take a few minutes to call out Cisco’s “Migrating Workloads from Cisco HyperFlex to Cisco Compute Hyperconverged Solution with Nutanix White Paper” which really needs a tl;dr. And here I am, the rambling asshole, hypocritically complaining about Cisco rambling on.

Moving from HyperFlex to Nutanix isn’t fundamentally different from other migrations…

  • If you’re moving to VMware on Nutanix, just set up a separate cluster and do a vMotion.
  • If you want to move to AHV, or have network considerations preventing cluster integration and vMotions, use Nutanix Move.
  • If you have a third-party backup/replication solution in your environment, which support the source and destination solutions, and are feeling spicy, that’s also an option. Cisco’s doc calls out Cohesity and Veeam specifically as they’re core Cisco partners.

In other words, if you’ve been through migrations to HCI then it’s largely more of the same. There are a lot of considerations in the whitepaper which are absolutely worth a review such as capability comparisons and Cisco nonchalantly pointing out how simpler Nutanix’s network stack is. You just have to bear with the authors writing out “Cisco Compute Hyperconverged with Nutanix” 158 bloody times.

Conclusion

If you made it this far, and yet another one of my rambling journalist entries, please do me a favor and let me know how many damn typos I need to go back and fix. If you still have any questions or concerns about all this reach out to me or your trusted technology partner. There’s no reason to freak out now but it can’t hurt to start having some conversations.

 

Additional Resources

Publication History

  • Oct 6, 2023 – Initial Posting