Do others sense this? If so, what options do you see for folks to keep their servers but move off of VMware? Is it all RedHat?
HPE did a big brain move to support multiple hypervisor backends with their own frontend. The only way to go forward imho.
I'm using Proxmox at my current $dayjob, and we're quite happy with it. I come from a big VMware shop and I think most businesses could easily replace VMware with Proxmox.
I think Proxmox should just launch an Enterprise contract, regardless of the cost, just have one. Because right now I think the main obstacle halting adoption is their lack of any Enterprise SLA.
On a personal level I would love to see KubeVirt, or Openshift with KubeVirt, take over more. It just seems like a genius move to use the already established APIs of kubernetes with a hypervisor runtime.
There's also Harvester "open source hyperconverged infrastructure" https://harvesterhci.io/
Or some Xen spinoff like https://xcp-ng.org/
Smaller shops are migrating to Proxmox.
Proxmox may come to many an HN visitor's mind (and I use it myself extensively, all my home services run on it), but it actually doesn't have a lot of enterprise features and isn't a drop-in replacement.
Not kidding, that’s the main blocker. We have the DevOps knowledge on our team to go to containers, prepackaged dev environments, etc. But corporate cyber tends to respond to our requests to discuss cyber policy and escalate via proper channels with “sorry that’s against policy”.
This is not my experience at one company but multiple good, name brand companies that generally do good engineering and software work.
NVidia are pushing hard in the direction of combined accelerators and ARM CPU (i.e. DGX, Thor, Jetson, etc).
Some of the upcoming hardware hits a sweet spot in terms of performance / $ / W. It's hard to ignore.
But Proxmox is ignoring ARM. Which is a big mistake IMO
They went to Nutanix right before the broadcom acquisition and never looked back.
They were much happier, and HCI was very nice for k8s nodes.
"Proxmox VE: Import Wizard for Migrating VMware ESXi VMs", 100 comments (2024), https://news.ycombinator.com/item?id=39841363
(I work for one.)
If you are on a perpetual license you can put the management vlan on a network not connected to the internet if it wasn't already and realistically this buys a few years. You will not be able to patch, eventually auditors will not accept that. For the rest not on perpetual licensing, when the licensing expires you will not be able to power on machines, if they reboot they stay off.
About half of clients we are migrating to hyper-v. Most are already running windows servers. There are some differences but hyperv covers the important features and the licensing is basically already included. Beeam makes the virtual to virtual move a lot easier, this is what most of our customers use for backups
For a good chunk they are migrating to azure or another hosted environment. If you don't have a main office with a file server or some more demanding line of business apps this is a pretty easy move.
A few are going to nutanix. Or more of expanding nutanix.
You dont think enterprise IT does sensible things like have multiple vendors to avoid single points of failure.
Our Hyper-V environment came online a few months ago. It was already included with our ELA with Microsoft so we were able to splash out a bit for some higher tier support.
Granted, we have a separate team working on "genAI stuff."
We started converting virtual machines about 3 weeks ago and we've gotten through ~500 of about 3500 or so.
Our grant based HPC environment is just moving back to bare metal. The VM conversion is just for ad-hoc HPC and then all of our general infrastructure. Some of our larger application servers (SAP Hana) are possibly staying on VMWare if SAP won't support them on Hyper-V.
This summer sucked big time but we'll make it.
Kind of sad seeing businesses getting screwed by closed source proprietary software, then making the same choices all over again.
Nutanix also seeing huge demand.
Not everyone is repeating their mistakes, with Proxmox and Xcp-ng seeing huge new level of business, as well, which is nice.
I'm part of the Apache CloudStack project and that too is seeing unparalleled levels of demand. The KVM hypervisor has sort of become the de facto choice, thanks to virt-v2v tool which can help migrate VMware guests.
Virtualbox
Parallel
Hyper-V
Heard from a peer school on the east coast that had to sign a 600% hike in their most recent contract. Absolutely evil.
If I were running the migration, my preferred pathway would’ve been to Apache Cloudstack. We had the expertise to pull it off, and it would’ve freed us from vendor partners. Nutanix was really only on the list purely from its technology portfolio; its lack of profitability and shifting towards SaaS for features like cost analysis meant that we’d be moving into a similarly bad situation as VMware at the time (wholly beholden to their business priorities instead of our own), which I didn’t care for.
There’s a lot of options out there, most of which are built atop either KVM/QEMU or OpenStack. Virtuozzo’s offerings impressed me, but the lack of a “comprehensive” product was a turnoff. Oxide was incredibly interesting from a simplicity and integration perspective, but the appetite wasn’t there to try a startup’s product. Microsoft and Oracle were both ruled out due to higher costs and more onerous licensing than VMware/Broadcom, while IBM/OpenShift were ruled out as our private cloud estate was 100% VMs with only ~20% of our internal products capable of containerization support.
The biggest advice I can give is to understand your workload today, and determine options accordingly. Everyone’s pitching K8s and containers, but if your estate is majority VMs, then a lot of those options just aren’t worthwhile.
We're driving anything that cannot be containerized to lift and shift to IAAS and forcing the app owners to pay for it out of their budget as motivation to modernize. They have to explain to the board why their spending increased and they are still on legacy.
- [1] - https://www.talos.dev/
Nutanix is just as expensive and also a locked in option.
I am seeing Nutanix the most, then Proxmox, Openshift.
For some sub products, Avi is often going to HAproxy, Aria to a combination of Terraform, Datadog (and others)
In theory, it's great. In practice, if you need to get "support" from someone else, it's not so great anymore, as all these companies have been discovering.
I would use VM technology if whoever wrote it would provide me with a contract saying that if anyone were to find just one program that would crash their VM (while not crashing a real machine) or miscompute, that I would get a billion dollars.
To answer your question: I was smart enough to never use it in the first place.
Some shops here migrate to proxmox as a UI because of certification requirements, but I migrated some of my customers to cockpit dashboard, and some to kubernetes. It's always a matter of scale and provisioning requirements.
Cockpit is my favorite so far because it's easy to setup, but its focus isn't cluster scale, which is what most larger companies need. You have to setup basically two cockpit variants: the webui and lots of cockpit server daemons (aka libvirtd on remote machines). The webui then uses SSH to login to other machines to manage them (e.g. via the known_hosts file on the webui server). [3]
Proxmox is pretty old and Perl, but it's doable. Usually storage clustering is a bit painful because you need something on a filesystem layer like ceph clusters.
There's also openshift but no idea if that is an IBM/RedHat lock-in as well, so the SMEs didn't want that risk.
[1] https://cockpit-project.org/
[2] https://www.proxmox.com/en/
[3] https://cockpit-project.org/guide/latest/feature-machines.ht...
Hire a couple of sysadmins who know their ass from a hole in the ground.
For Windows shops (you sick bstrds :D) -> Hyper-V
RE: Proxmox, not a hater, but errtime I read that name, I picture "Mike Myers Dieter". Seriously tho, the best they could do for themselves would be to make ProxMox into a UI for the Incus REST API (and their legacy backend) then repackage their enterprise offerings as add-ons for Incus (for email, DRS, etc.) BTW, the first person to release an "I use this exclusively everyday for months before releasing it on Github" gesture-controlled WebGL/Audio and WebXR based UX with agentic Incus REST API sensor analysis and settings controls, will win a valuable prize.
https://aws.plainenglish.io/why-even-google-is-rethinking-ku...
-> Mind earlier warnings that many people in the Industry these days tend to follow each other around on forums creating a sort of "feedback loop" so bad trends (eg, PHP) takeover the game.