Broadcom Ends Support For Free ESXi Vmware Hypervisor 105
stikves writes: Today, Broadcom announced immediate end of ESXi availability. ESXi has been an important tool for many "homelab" enthusiasts -- offering simple bare metal virtualization for small setups. Unfortunately they don't offer a replacement, except for paid subscription services.
Nail in the coffin for Vmware (Score:5, Insightful)
In 10 years it will have been replaced by something else. Mark my words.
Re: (Score:3)
Re:Nail in the coffin for Vmware (Score:5, Informative)
I'd say Xen has largely gone away for 'new adoption' as well. The dom0 concept meant your 'host OS' was almost, but not quite your host OS, and while Hyper-V has a similar architecture, it's less broken about it. Every Linux-leaning shop I've seen goes qemu-kvm as the virtualization/emulation approach nowadays.
But either way you slice it, I agree that VMWare's customer base are largely businesses that decided to do x86 virtualization about 20 years ago and have never taken a pause to evaluate alternatives as they became available and matured.
Re:Nail in the coffin for Vmware (Score:5, Insightful)
This is just more value extraction. There's no need to maintain the onramp when you've already got your customers.
"Krause told investors that the company actively pursues 600 customers -- the top three tiers of the pyramid above -- because they are often in highly regulated industries, therefore risk-averse, and unlikely to change suppliers."
https://www.theregister.com/20... [theregister.com]
Re:Nail in the coffin for Vmware (Score:5, Informative)
Exactly, and this isn't the first time that exact pyramid appeared in a Broadcom acquisition strategy either. They are explicitly happy to lose 100,000 accounts in the name of gouging the hell out of 600.
Re: Nail in the coffin for Vmware (Score:2)
Re: (Score:2)
Re: Nail in the coffin for Vmware (Score:2)
Re: (Score:2)
I understand that the plan is that those 600 customers will be willing to pay more since they are "locked in" for the forseeable future and that means they can cut developer time / support to just focus on the needs of those few customers.
Meaning they don't have to bother about working on features needed by the other orgs / smaller businesses, or have enough support staff to support the 10000s of smaller customers.
Presumably they will be able to cut alot of cost and milk those few customers and still profit
Re: (Score:2)
Exactly, and this isn't the first time that exact pyramid appeared in a Broadcom acquisition strategy either. They are explicitly happy to lose 100,000 accounts in the name of gouging the hell out of 600.
Capital is rent-seeking again I see. I guess I shouldn't be surprised that there are no laws surrounding such behavior. The behavior destroys industries but is extremely profitable.
Fucking lazy parasites.
Re: (Score:2)
This is just more value extraction. There's no need to maintain the onramp when you've already got your customers.
Typical short-term (and mediocre CEO) thinking. A few years down the road, the next CEO will have to explain to the shareholders why the company hasn't been growing and how he's going to remedy that.
Re: (Score:2)
The operational word is 'the next CEO', so the current CEO (who incidentally doesn't give a crap about the business beyond his next stock vesting) makes out like a bandit while burning the business to the ground.
So that CEO would take issue with 'mediocre', from their perspective it all worked out amazing.
Re: (Score:3)
I am not sure Broadcom is taking the wrong strategy here.
Virtualization is a commodity product now. No libvirt+qemu+kvm isnt a drop in replacement for VSphere, but what is missing is mostly glue, not 'high-tech' stuff, in other words you can probably contract for a few devs to build whatever specifically you need.
Continued development of ESX probably represents a lot of cost for a market segment that is only going to get smaller, no matter what you do. Even if there is a move from the cloud back to the corp
Re: (Score:3)
Keep in mind that without a competitor, Hyper-V won't get much love, like File and Print Services for Netware (FPNW), Microsoft's Netware killer, kind of died on the vine after 2000 or so.
Install NT 4.0, license 426-1111111 and then put FPNW atop it and you had the functional equivalent of any Netware 3.11/3.12 server with a size only limited by available bandwidth and compute. No wonder Novell died. Same trick MS was pulling with Hyper-V.
Re: (Score:2)
Frankly, does Hyper-V need 'much love'? It's certain they'll keep it up to date with the platforms which is necessary, beyond that it is pretty much 'done'.
It will continue to face competition from the likes of Proxmox and RedHat.
Frankly, vSphere hasn't changed anything fundamental much in years, despite some version bumps. They painted themselves into a corner with no logical place to go (all the cool stuff was on top of actual operating systems, and they were very much trying explicitly *not* to be an op
Re: (Score:2)
I imagine you weren't at Ready in Vegas in 2019 when all the fist pumping and bellowing - full monkey boy dance mode - about getting VMware running on Azure. Sort of. Anywho, MSFT thinks VMware as a gateway to Azure is a big deal and competing with Hyper-V was important from that perspective. Absent a real competitor - c'mon, Proxmox doesn't even figure on their radar - they won't even bother. They seriously considered making 2019 their last version of Server.
On-prem infrastructure at MSFT is getting li
Re: (Score:2)
They seriously considered making 2019 their last version of Server.
Was this last version as in "Windows 10 will be the last version of Windows, it's now a rolling release" or last version as in actually discontinuing the product?
I would say that even if it were the latter, competing with VMWare wouldn't make them work *harder* on their on-premise solution. Microsoft has been full tilt burning any on-premise to the ground, because it's just not as profitable as holding the customers hostage to a subscription.
Re: (Score:2)
I wasnt but of course MS wanted VMWare running in Azure - that is obvious path as far as lift and shift migration of on premises to the cloud. However beyond - it works and you can manage and support it organizationally 'well' enough to allow you to migrate those tricky hosts, what more is needed? The point is allow places that are seeking to go 100% cloud to get that done more quickly so companies can shed the local DC costs sooner and start renting compute time from M$ sooner.
Long term how "good" VMWare
Re: Nail in the coffin for Vmware (Score:1)
Plenty of glue is out there. Nutanix, Proxmox, XCP-NG, Ubuntu, Red Hat, pretty much everyone has feature-complete, near drop-in replacement for whatever you want in VMware for a fraction of the cost.
You can migrate about any VMware infrastructure without even needing to buy new hardware.
Re: (Score:1)
Re: (Score:1)
Well, perhaps you should kick all your proprietary vendors to the curb, wouldn't be a bad idea, what happens to your backups when Nakivo (never heard of them FYI) goes belly-up because their VMware customers abandoned them?
Nutanix is KVM based so if something can backup Nutanix, it can backup most of the other ones with perhaps the exception of Xen-based stuff like XCP although Citrix isn't exactly an unknown company to the big enterprises either.
If your backup program can't target a generic storage and res
Re: Nail in the coffin for Vmware (Score:2)
Re: (Score:3)
Typical short-term (and mediocre CEO) thinking. A few years down the road, the next CEO will have to explain to the shareholders why the company hasn't been growing and how he's going to remedy that.
This seems unlikely; "VMWare" isn't "the company" anymore, it's Broadcom, and Broadcom has been following this strategy for some time now. Its shareholders seem happy with the way things are going--their stock price is 5x what it was four years ago, and >20x what it was ten years ago. Revenue is up 14x over that same period.
Personally, I hate these sonsofbitches and what they do to the companies they acquire and their acquisition of VMWare has certainly screwed me in the mid to long term, but the share
Re: (Score:2)
The only problem is that virtualization isn't your normal Enterprise Lock-In. It's literally the abstraction that prevents it in the first place! VMware will be dead within five years, probably much shorter.
Re: (Score:2)
You are thinking too logically. Logic has no place here, this is business IT.
Some people were *convinced* that software couldn't run under virtualization versus baremetal unless the software vendor supported it.
Now you have software which 'certifies' that it works under vmware. CIOs eat that up.
Re: (Score:2)
Nah - if you use ESX / VSphere - things are highly coupled to your backup plans/software/equipment, hot/cold site fail over, storage design/controller choices etc.
Is it hard like swapping out your ERP systems; but its non-trivial.
Re: (Score:2)
But either way you slice it, I agree that VMWare's customer base are largely businesses that decided to do x86 virtualization about 20 years ago and have never taken a pause to evaluate alternatives as they became available and matured.
I spent almost 14 years recently working for a Fortune 500 company. My severance package doesn't allow me to name them - yet. But I can tell you that they often made decisions on finding another company to blame if stuff went wrong. For example, my department used completely free Linux versions for years but eventually they made us switch to a commercial variant so we had someone to blame if anything went wrong with it. When I was last there, we were starting to move a few customers to AWS instead o
Re: (Score:2)
Xen hasn't been a serious part of the picture for most considerations for close to a decade, perhaps longer, due to the overall complexity and inflexibility of deployment. It's heavy coupling to storage location/container is a particular agitation...
Re: (Score:2)
I can only hope.
I frankly thought the exact same thing 10 years ago. At the time Hyper-V and QEMU-KVM had already had 5 years under their belt and I was sure the market would shrug off vmmware and go to their respective corners where 'virtualization' is no longer a wholly separate thing and just becomes part of the 'operating system' management you already did. Whether your "main" workload was more Microsoft-like or Linux-like, there was a virtualization solution more affine. By it's nature, no one could
Re: (Score:1)
Re: (Score:2)
Did it a long time ago. Switched over to Proxmox, a web-gui for qemu with a bunch of really nice features, backup solutions, container support out of the box and a bunch of other stuff.
Of course they try to push you towards a subscription, but their free tier is just as good, if you're willing to spend time on their boards instead of asking their support for solutions. The usual spiel, if you're a company and an hour of time costs more than 100 bucks, you're better off with the sub.
Re: (Score:2)
You are aware that free or pirated software is the way students get exposure right?
Re: (Score:2)
Not always. I know our students get a license for most everything from Microsoft that is valid as long as they are our "active" students (2 major terms in a row not being registered, summer term isn't considered major just fall and spring).
I also know I've looked into various other software packages for lab use and most of the vendors I communicated with indicated they'd give student free licenses for home use if we had a lab full of licensed installs (this was for some CAD/CAM stuff, etc).
Re: (Score:2)
Because enthusiasts who have home labs carry that skill over into businesses.
The choice in our environment now was actually ESXi or Proxmox, because those were the two solutions our team has experience with. Why? Because people use it at home. We're lucky the coin toss came up Proxmox.
Re: (Score:2)
Are they purposefully killing it? (Score:1)
Every single decision Broadscum has made seems to be one of purposefully killing a product. I don't get it.
Re: (Score:3)
Broadly speaking, they explicitly have a strategy around "if we lose 80% of our customer/user base, it's fine as long as it lets us gouge the remaining 20% by at least 500%". Correlated with a presumed plan to have cashed in their restricted stock and go on to next company/retire by the time consequences ensue.
Incidentally, if you look at 'enterprise' IT, you can see that it can take over a decade for consequences to catch up, so it's a pretty safe bet you can gouge the hell out of those shops until your s
Re: (Score:3)
I have been in a company with that strategy. About 80% of their revenue came from a single customer.
What do you think happened when that customer noticed they got the company by the balls?
Re: (Score:2)
Their goal is to jackpot the customers that can't move, which is extremely stupid once you realize we're talking about a stack of products that make it easy for customers to move, and your competitive advantage was that you made management of everything a little easier.
Re: (Score:2)
The product is ALREADY terminally ill. They're just extracting as much as they can before it becomes unsellable.
VMware was a leader by staying ahead of the curve. That was their value.
They virtualized compute.
They virtualized storage.
They virtualized networking.
There is no where else to go and competitors are slowly catching up. Once competitors meet 90% feature parity there is no justification in budgets for VMware.
VMware shareholders know this and don't want to be holding the bad. In comes Broadcom to
No big surprise, just disappointment (Score:3)
Re: No big surprise, just disappointment (Score:5, Interesting)
Funnily enough, id say ceph is a passable "filesystem", but it's first and foremost an object store, particularly well suited at backing "blobs" like virtual disk images.
ISCSI snapshotting isn't really an iscsi feature, it is really an integration with your storage provider. If, for example, NetApp, you are really asking ON TAP to internally clone and manifest as new ISCSI target. Incidentally, that snapshot is ultimately a file in WAFL, which is a filesystem.
Peel the block storage onion in a modern storage architecture back enough and you'll find either a "filesystem" or at least an object store underneath. I suppose LVM offers a more old fashioned facility for snapshots, but generally people don't bank too hard on it because it's just not as flexible as other approaches.
Re: (Score:2)
Note that the industry has invested a great deal in making block oriented storage seem 'magic' and a solution that manifests NVMoF or iSCSI is 'obviously' better because it's "block" all the way through (it won't be, but a high cost storage vendor will do it's best to bury that abstraction out of sight to encourage the perception). It shouldn't be too surprising when someone has bought into that story with so much marketing push behind it.
Of course, funnily enough, iSCSI is actually a terrible abstraction
Re: (Score:2)
I may be wrong but I think that was added a few months ago. But maybe I mistake it for something else, I don't have that problem.
Re: No big surprise, just disappointment (Score:2)
XCP-ng looks like a great replacement, with a decent web interface and everything.
Re: (Score:3)
We use XCP-ng in the office, but it has one major flaw. You can't nest M$ Hyper-V in it. The M$ Hyper-V fails under XCP-ng, but seems to work in other nest virtualization setups. Long running bug that really needs fixing, as it prevents being able to run WSL2 on virtualized workstations. And I understand it causes issues with Docker on windows as well.
https://github.com/xcp-ng/xcp/... [github.com]
Re: (Score:1)
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Works in a pro environment too, unless you want to manage a few 100 metals running a few 10,000 VMs and load balance them, that's a wee bit outside its paygrade.
But frankly, if you can afford some 100 blades, don't be a cheapskate and get a decent virtual manager that can handle that load.
Re: (Score:2)
Honestly, your UID is low enough that the only person to blame on this is you.
The writing has been on the wall for VMware for over a decade at this point: that it took Broadcom to acquire them to push folks out is just a detail. The product has been increasingly verticalized, with ever-increasing Oracle-like licensing and hostility towards technical early adopters.
The proxmox documentation is quite good and thorough. Arguably, it's easier to deal with than VMware's documentation, if you're comfortable with
Re: (Score:2)
Re: (Score:3)
Re: (Score:2)
Oh - only a hobby! Totally excused, then. My apologies for coming off judgy.
ESXi has been "dead" from a hobbyist POV since 6.5 (or whatever it was) lost its hardware support, way back around 2012. That's when I'd decided it wasn't suitable for further deployments. If I have to pay for test systems, I don't want it.
Re: (Score:2)
Exactly. Gave up on ESXi in my home lab around the time 7.x was introduced and VMWare failed to support a lot of older hardware (eg RAID or HBA controllers) that home labs are often built around. Proxmox has been great and offers anything I've needed for clustering and HA.
Re: (Score:1)
> Honestly, your UID is low enough that the only person to blame on this is you.
and this is my second UID too! You're not wrong. I've been thinking about getting off ESXi for years, but it was so stable, and it's a hobby, not a job, so I did other things to my setup. I had been making preparations, but not very quickly.
I'll use the time to check out some alternatives, I'm installing xcp-ng as I speak, and Proxmox looked fine anyway - just wanted to make sure I knew enough to know what I liked before I jumped ship.
Your low uid comments made me laugh ;) It's both a blessing and curse - get off your ass and lay down that low uid knowledge.
Re: (Score:1)
Your low uid comments made me laugh ;) It's both a blessing and curse - get off your ass and lay down that low uid knowledge.
The only true low UID knowledge is knowing that saying nothing is often the best choice.
Which means I just screwed up.
Re: (Score:1)
I fully agree :-)
Re: (Score:1)
Proxmox is based on debian; as long as you don't mind relatively stable featureset - you will be pleased. Promox is not an enterprise replacement for vmware, as it doesn't have all of the concepts of scaled capacity pooling and antiaffinity (yet); but for a home lab setup - you can easily run up to a hundred smallish vms on a 4-5 node cluster and be pleased. It's a great dev / test environment, and can do prod to some defintion of scale. The more you can build from bare metal up, the better you can scale
Oh well (Score:5, Interesting)
Proxmox is free. Windows Server Hyper-V is (surprisingly enough) free as well. I haven't used ESXi in ages, but last time I did, I remember that I had to do a whole bunch of extra work because X or Y key feature found in Vsphere wasn't supported on the freebie version.
The world doesn't explicitly need ESXi any longer and it sounds like the move to cater to Broadcom's biggest customers will just lead to a world where fewer professionals are familiar with its platform. Good luck with that.
Re:Oh well (Score:5, Informative)
Windows Server Hyper-V is (surprisingly enough) free as well.
The free standalone "Windows Hyper-V Server" is dead (2019 was the last version). The Hyper-V role is still there (and will continue to still be there) but is licensed as part of the Windows Server OS so is not, strictly speaking, "free."
Re: (Score:2)
Windows Server Hyper-V is (surprisingly enough) free as well.
The free standalone "Windows Hyper-V Server" is dead (2019 was the last version). The Hyper-V role is still there (and will continue to still be there) but is licensed as part of the Windows Server OS so is not, strictly speaking, "free."
While MS is no longer releasing new standalone Hyper-V servers, they've stated that the 2019 version will remain available for download and will be supported until 2029. So if you're wanting to dip your toes into a Microsoft-centric virtual server setup, that's still a cheap way to do it for another 5 years. Plenty of time to evaluate your options and decide if it's worth moving on to a licensed copy of Windows Server (which as you noted, always has the Hyper-V role available).
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
it sounds like the move to cater to Broadcom's biggest customers will ...
it sounds like the move to bleed Broadcom's biggest customers will ...
VMware player is coming up next (Score:1)
Never liked it (Score:2)
Although heretical, I host all my Linux VMs on Hyper-V
Re: (Score:2)
Agreed. I would not be surprised if that was an intentional design decision on their part, though.
Seems like they want to kill it (Score:2)
I'm good for now (Score:2)
I'm already frozen on ESXi 6.5 because of compatibility issues they started introducing after that. I have an iso with all the NIC drivers they pulled after 6.0 added.
I'll likely just stick with it until there is something I need to do that it won't, but I don't see that happening until my next hardware refresh... Not much you can do if some new driver is required and support is gone. Then I guess it'll be ProxMox.
Proxmox (Score:5, Interesting)
Proxmox is free and stable with great management tools.
It also has Proxmox backup server which has time compaction and file level restore.
We run about 60 VMs with 6 servers and 2 SANs, have dashboard monitoring for it all. Been rock solid.
Can I buy a license? How much? Alternatives? (Score:1)
I was just thinking of building a new multi-head desktop like that described here: https://www.pugetsystems.com/l... [pugetsystems.com]
The example from Puget Systems is for gaming but I'm not necessarily concerned about gaming but just getting two or three nice workstations in one box. If ESXi isn't free then how much to buy a license? Are there other hypervisors that offer similar features? The one feature that made ESXi suited to this multi-head style system was PCI pass-through. I want the VM to have direct control of
Re: (Score:1)
Your article is a decade out of date there buddy
I know it's old but it's an example of the general concept, not something I'd try today exactly as described.
Re: (Score:2)
You can do PCI-Passthru with Qemu-KVM/ with or without libvirt.
They seem to have used dedicated GPUs for each head. So you can still build something like this with the host being any light-weight Linux distribution really.
About the only thing FOSS does not have is a good V-GPU solution with a windows driver. So if you want a windows guest with GPU support you'll have to give exclusive use of physical GPU. Virtual GL works for other Linux guests, but I don't know if the performance is all that good or not b
Re: (Score:2)
Had to do some of that a little while back; and found that getting anyone to confirm the presence or absence of PCIe ACS was like pulling teeth; and that there were PCIe periphera
Re: (Score:1)
Basically all the hypervisors support PCIe passthrough(except the 'desktop' ones, neither vmware workstation nor win10/11 hyper-v do); though there's a risk of...complications...because doing that relies on the platform's IOMMU and PCIe ACS support to both exist and not just be a buggy stub that's enough to tick some checkboxes.
Which of the hypervisor options handled PCI pass-through as easily and stable as ESXi? I like ESXi because it did PCI and USB pass-through so well. USB pass-through might be enough for me to get what I want but I'd prefer PCI pass-through for simplifying the handling of multiple displays. PCI pass-through on a USB controller means I can skip a step on directing devices to the intended VM by just picking the appropriate USB port to plug in the device. I'd miss that if I can't get that. I've also become
Never assume... (Score:2)
...that a free and useful piece of software will remain free
Mercenary weasels are always on the lookout for free stuff that they can turn into a subscription
The enshitification continues
Sigh (Score:2)
Re: Sigh (Score:2)
End of an era... but unsurprising (Score:2)
I'm sorry to see this go. I have fond memories of deploying the free version of ESXi for situations where there was just no budget for a commercial alternative, and the goal was simply consolidating 4 or 5 physical rack or tower servers into one machine to save on electricity, etc.
But I know time marches on, and there's not much use-case for NEW customers to pay for VMWare ESXi. So Broadcom has no real incentive to maintain support for the free edition. At this point, ESXi is mostly about existing customers
XCP/XO + more (Score:2)
Basing your operations completely on proprietary solutions is just asking for being extorted later. In this realm, fortunately, there are FOSS options. And options that are not only very robust, but most also have commercial support options as well.
Knowing your pricing is going to increase a lot, many companies would do well to swallow some pain now and shift to FOSS platforms, then later save tons of money on licensing and with much lower-priced support options. And with more money flowing into such projec
Re: (Score:2)
Re: (Score:1)
my last job used asterisk, the owner and I both knew how to configure it, it's not rocket science. worked great with Cisco phones (and polycom) even! Certification for what?
Cisco is the rube goldberg bespoke solution at five times the price. ditto for their routers, blades, switches....
Harvester, Proxmox & Nutantix for the win (Score:2)
broadcom is boeing (Score:3)
Broadcom is akin to Boeing, a place where R&D is non-existent, where support is a joke, where issues are ignored, where development stops.
Everything they touch eventually turns to ash. They only care about value extraction short term, and that milk that cow to starvation.
VMWare is done.