VMware transition to subscription, end of sale of perpetual licenses
news.vmware.com
news.vmware.com
Then I got to see the conversation about subscriptions at VMware. The storyline was always “wouldn’t it be great for us if we switched to subscriptions? Look how good all the math gets!” The customer barely appeared as a background character.
Especially entertaining when it came to things like PWS. After the acquisition VMware shut it down — because it didn’t have any real customers — and then in the push to subscriptions the new folks in charge of TAS were like “Great idea everyone! What if we provided Cloud Foundry… as a Service!” The engineers who had been around long enough to see the whole cycle tried to explain but I don’t think it got through.
This is described in the book as a fantastic achievement for Adobe but one that was quite difficult to get off the ground. On reflection I do not share that assessment and while it may have been personally successful for the CEO/author of the book, it was done entirely at the expense of the customer and their experience. I would in fact consider renting software that takes up storage on your own hardware to be anti-consumer.
this is what it means to be a monopoly.
Adobe may not be one by the legal definition, but it certainly is a monopolistic suit of products, as their "customer" does not have a way to switch, regardless of other competing software.
I personally think that the current legal idea of a monopoly is too narrow. I say the definition should be focused on the customer's cost of switching - and if it's higher than some reasonable threshold, the provider of said service/product must be considered a monopoly.
The best story I heard on this was from Zappos and Tony Hsieh. Zappos was famously customer focused (which probably played a big roll in Amazon acquiring them). The story goes that after a night of drinking they had someone call Zappos to find pizza for them, and Zappos customers services helped them out (not knowing the CEO was there listening).
Did you hear about the CEO that told institutional investors that the company missed projections, but that they shouldn't worry about it because he'd help them order a pizza?
Me neither, but it shows how absurd this sort of story is if the shoe were on the other foot.
There are outlets for cheap shoes (in this case) without overhead from support and other sources.
If you have a 50% profit margin, you're not sensitive to small (or even medium) fluctuations.
That's not a universal. I can give a half-dozen counterexamples (including Walmart) which, for example, operate with extreme efficiency, or achieve massive scale. But it's a good rule-of-thumb.
Running a business, there are sometimes more important things than profit. Robust businesses are more pleasant to run and to work in than ones with razer-thin margins (even if bigger and more profitable). As an entrepreneur, if I were to pick:
1) Guaranteed income of $200k for the rest of my life (inflation-adjusted) with fun work and reasonable work-life balance
2) A $20M payout with 50% odds
I think I'd pick #1.
Ergo, we see a lot of lifestyle businesses, businesses which don't try to grow, or otherwise. Yes, one can make a lot of arguments about fiduciary duty, shareholders, etc. Some of those are even correct for a large, public corporation. But most don't apply to a founder-owned founder-run business.
Costco shareholders were very vocal with the founder of Costco, wanting him to do things that were against the mission of the company, all in the name of making a quick extra buck in the short term. He told them to get bent, and Costco has a lot of die hard fans who will defend Costco until their dying breath. You can't buy that kind of loyalty with some extra profit after treating customers like shit.
And obviously, if a company is going to spend extra for excellent customer service, the business needs to be able to support that. Operating at a loss, thinking they'll make it up on customer service, is not a business plan. That isn't what we're talking about here.
JetBrains, for instance, also does subscriptions. They are generally lauded as doing them right. I would never compare JB to Amazon or VMWare just because they both do subscriptions.
Therefore jetbrain cannot afford to alienate their customer base. Unlike vmware or adobe.
The challenge is that companies architect their services and applications around what they have, and they may be very operationally dependent on specific features of vCenter and how they've configured those features.
If you want a deployment choose something like openstack-ansible.
If you know (say) Ubuntu and Ansible, then there's pretty good automation with openstack-ansible and kolla-ansible. There are a number of moving parts, but an intermediate-level Linux sysadmin can get things going without too much difficulty. There are vendor-provided solutions (e.g., from RH) as well if you don't want to do a fully DIY, or want commercial support.
* https://www.openstack.org/marketplace/distros/
At a previous job we had a team of 4-5, and 1.5 FTEs split their time between our HPC cluster and our OpenStack cluster (20+ hypervisors, ~10 Ceph servers; >350 VMs/instances).
* https://www.vmware.com/ca/topics/glossary/content/openstack-...
the point of vmware is that is fairly trivial to go from a single server with some VMs, to a massive cluster with live zero-downtime migrations. (and have funky storage, get working snapshots and backups.)
Sure openstack can do that. but its non trivial to setup.
I've seen a very small outfit spend six months with all hired consultants to set up a 3-server VMWare cluster.
VMWare doesn't seem that trivial.
Then those people are woefully incompetent and will be regardless of the software they're using.
VMware (wasnt) that hard. How many VMs do you need? work out the storage requirements, triple it, plumb it in, get the right licences, and go.
That’s basically what stopped it back in ‘14 or so. It was too hard to provide software that could run on an OpenStack because it was such a choose-your-own-adventure.
What's next, a survey of people attending the annual dentistry conference shows that they think dental health is important?
Large and Enterprise companies won't really be bothered by the change except the price, and unless other platforms include all the "Enterprise Check-boxes" they think they want, I'm not sure how fast adoption will be if at all for stacks like Proxmox.
I am not defending the business model or the tech offerings of Vmware, Microsoft, etc -- far from it. They're extremely bloated, buggy, and unless you happen to be one of VMware or Microsoft's most expensive customers, the support might as well just be a sin tax. Even for the "biggest" clients, they were not able to get Vmware or Microsoft to budge on some pretty obvious and complete blocker bugs that would drop production environments entirely.
Enterprise is weird -- I would be extremely happy if not a bit exhausted supporting a large OSS virtualization infrastructure for clients, but I just do not see it happening without a lot of "carrots" for such companies from the virtualization vendor. Enterprise business is basically about shifting responsibility and liability for your tech-stack to someone else -- yes, it requires strong tech skills to use these tech-stacks, but the more maintenance/liability that can be shifted to the vendor, the more preferable it is for most companies.
I don't know what the businesses will choose -- I'm hopeful medium and smaller businesses will shift to more open solutions and from this business, the open solutions will be able to grow and become more attractive choices for enterprise, I'm just not sure that such solutions will avoid the tech industry's obsession with subscription licensing and feature bloat to check boxes.
From what I've seen none of them have taken this issue seriously, either, and won't until they get squeezed hard enough, at which point they'll have to... spend five years trying to move!
Keep in mind it is more than just shifting all of the VM workloads to a new platform something not easy in the first place
But you also have to factor admin training, backup and recovery systems, monitoriting systems, automation tooling, etc.
We have been using vmware since version 3 or 4, our entire team is almost exclusively training on vmware, with limited knowledge on hyperv and virtually none on anything else
Something like ProxMox that has no third party support, no veeam support, limited automation tooling, limited support for commerical monitoring (like PRTG) etc etc is probally a non-starter unless those things support it.
It would cost us far far far far more than 3x licensing increases to git rid vmware. Broadcom knows this.
Honestly VMWare has not been increasing its prices as aggressievly as other vendors anyway, of my software licensing fees I pay annually vmware is one of the cheaper vendors I have, not even breaking into the top 10, I pay more for Veeam, Windows Server, and many many many other things than i do vmware
Proxmox VE has an extensive REST API with API tokens and an access control system to allow for safe automation. There are also Terraform providers and Ansible modules for those that do not want to use the API directly, or use those automation tools already.
There's also Proxmox Backup Server that can handle guest-level backups as good as Veeam can, but yes, the agents and application level backups is still not there, but most of those are unrelated to the outer hypervisor, so one can still use their preferred backup solution for those parts just fine.
Central monitoring is possible through the external metric server integration.
Further third-party integration is a natural chicken-and-egg issue. For a lot of third party integrations you might just need to ask the vendor of said third party if they can support Proxmox VE, as long as they receive no inquiries as admins won't even ask if it would be possible, they have no incentive to add support for anything.
But it's getting better, and more eggs are hatching.
> It would cost us far far far far more than 3x licensing increases to git rid vmware. Broadcom knows this.
The migrations I saw in our enterprise support had way lower cost. IMO 3x the cost would need some heavily specialized workload (which IME fewer admins actually really have than they might think) and/or admins that are just heavily against any change (as yes, they must relearn how to do some things, there are training courses for a few thousand bucks that cover the whole stack though) or having an anti-FOSS/Linux'ish mindset. I.e., come to us with so many prejudices that it's hard to make them realize that, while there are surely things that can improve, that just the same on the other side of the "pasture", but overall the grass is quite green enough for most use cases already.
I am sure it does, that would then require enterprise to build their own tooling. I clearly talked about integration with 3rd parties that were built into the product, I do not know many IT Teams that have either the human resources to devote or in many cases the Knowledge to inhouse dev their one tools on that level. They purchase 3rd party tools in lieu of bespoke tooling.
This is a common point of contention when I interact with people in developer or devops communities like this one, they have forgotten how traditional IT shops work. We are still out here, and we are most likely the primarly customers impacted by this change. The DevOps shops are in the cloud...
>>>>There's also Proxmox Backup Server that can handle guest-level backups as good as Veeam can, but yes, the agents and application level backups is still not there
PBS is not a replacement for veeam even for Guest-Level Backups, It is about 5 years behind if not more in Backup Tech.
Veeam Provides (that ProxMox does not)
* Hypervisor agnostic backups (Restore from any supported Hypervisor to Any hypervisor or supported Cloud. i.e I can restore a vmWare VM to AWS Directly)
* Application Aware processing for common applications like SQL, Sharepoint, etc I think this is different than what you call "Application Level", as an Application Aware Backup will use technology to flush cache, finish tranactions, prune logs, etc to put the application in "saved" or "backup ready" state, PBS creates "Crash Consistent" backups only, basically as if you just pulled the power from the VM, some applications this will corrupt data, Application Aware backups work to prevent that corruption
* SQL Transaction Logs backup in per minute intervals with cross location functionality
* Continuous Replication
* Isolated / Automated Restore testing with reports
* Awesome Compression and Duplication rates in the backup files
* Builtin Support for S3 Storage with out having to do OS Level hacks (PBS depends on something like rclone to do this)
* Builtin Support for Block Replication with having to do OS Level Hacks
* Help Desk portal for File Level Restores that allow for RBAC Security
* Integrated Agent based backups for physical systems that provides a Single Plane of Glass for Backups
* Change Block Tracking (CBT) support
* Immutable backups repository
Comparing PBS to Veeam, is like Comparing pfsense to Palo Alto Pan-OS firewall, sure they both block traffic, but 1 in 100x more advanced
Besides the fact that I explicitly also talked about third-party tooling, which definitively is available, even if not yet as broadly (cue chicken and egg), lots of such extra third-party tooling makes one's infrastructure dependent by many more projects than required, each adding complexity and reducing reliability. But as said, this is a complaint for the third-party tooling provider.
> they have forgotten how traditional IT shops work. We are still out here, and we are most likely the primarly customers impacted by this change. The DevOps shops are in the cloud...
Yeah, we certainly did not forget that and cater that group extensively already. But currently the ones that use PVE or PBS are the ones favoring independence, i.e., FOSS and projects that only have to answer to their user, but no profit optimization machine, either by being bitten way to often by such changes already or understood early enough that having core infra on a proprietary platform including any 3rd party tooling is rather reckless for the long term (cue at this thread).
> Veeam Provides (that ProxMox does not)
s/ProxMox/Proxmox VE/
There's no upper case m, and Proxmox is the company, the actual products are Proxmox Virtual Environment (PVE), Proxmox Backup Server (PBS), Proxmox Mail Gateway (PMG), etc.
But, let's actually comment your points:
> * Hypervisor agnostic backups (Restore from any supported Hypervisor to Any hypervisor or supported Cloud. i.e I can restore a vmWare VM to AWS Directly)
AWS wasn't in our scope, but we can add any target that has an actual open specification, i.e., not requiring a deal with the devil that would force us to do a low quality hidden proprietary implementation.
> * Application Aware processing for common applications like SQL, Sharepoint, etc I think this is different than what you call "Application Level", as an Application Aware Backup will use technology to flush cache, finish tranactions, prune logs, etc to put the application in "saved" or "backup ready" state, PBS creates "Crash Consistent" backups only, basically as if you just pulled the power from the VM, some applications this will corrupt data, Application Aware backups work to prevent that corruption
> * Continuous Replication
Can be done by Proxmox VE's Ceph and ZFS.
> * Isolated / Automated Restore testing with reports
Wasn't done with PBS <-> PVE di
> * Awesome Compression and Duplication rates in the backup files
Yes, PBS has a content addressable storage with great deduplication and compression.
> * Builtin Support for S3 Storage with out having to do OS Level hacks (PBS depends on something like rclone to do this)
This is indeed lacking, but we're working on it, I'd think we have something read until mid of 2024.
> * Builtin Support for Block Replication with having to do OS Level Hacks
Not sure what you mean here exactly..
> * Help Desk portal for File Level Restores that allow for RBAC Security
We have a role based access control system for all our products, this can be already done...
> * Integrated Agent based backups for physical systems that provides a Single Plane of Glass for Backups
You can use the CLI client already, but yeah it should get some UX improvements for making that nicer OOTB.
> * Change Block Tracking (CBT) support
This is supported since years by Proxmox Backup Server. Sorry, but at this point I have to ask, did you even test Proxmox Backup Server or is your list just guessed, or partially AI generated?
> * Immutable backups repository
The PBS allows that too, we have explicit permission rights that can only add backups, never remove, plus on sync to off-site one can also disable any pruning, those features and our tape support makes it easy to have ransomware protected backups.
> Comparing PBS to Veeam, is like Comparing pfsense to Palo Alto Pan-OS firewall, sure they both block traffic, but 1 in 100x more advanced
As most your points above are already supported I'd take 1.5x at best, but thanks for your feedback, some of it was quite valuable.
for example
>We have a role based access control system for all our products, this can be already done.
So you have a web portal where a Help Desk Tech can take a user request for a file to be restored back to a vm, open a web portal search for the indivual file then either download that file and email it to the user, or restore it back to the VM, or restore it to a completely different VM depending on the use case???
Veeam Does
>>Can be done by Proxmox VE's Ceph and ZFS.
Sure you can do it that why, but veeam has support for this in their product, in the UI that is already part of my work flow, this is what I consider a "OS Level or Storage level hack" it is not really part of Proxmox, but part of the underlying linux OS or Storage solution.
>>Builtin Support for Block Replication
Veeam use ReFS or XFS to only store one version of a block no matter how many backups are taken, this is done in the veeam software and works with their VBK file format. ZFS has some dedup which is what proxmox relies on but it works differently than Veeam, and seems to be less effective. These Block Cloning also works with the Data Format uploaded to S3 to massively save on storage costs
>> I have to ask, did you even test Proxmox Backup Server or is your list just guessed, or partially AI generated?
About 3 years ago...
Have you ever used Veeam? I have been using Veeam since version 9.5, and hold a VMCE certification
> So you have a web portal where a Help Desk Tech can take a user request for a file to be restored back to a vm, open a web portal search for the indivual file then either download that file and email it to the user, or restore it back to the VM, or restore it to a completely different VM depending on the use case???
It surely isn't as integrated/polished, but that it's just not possible is far from true either. We're polishing release for release, most of those things are not complicated, but need a bit of time.
> Sure you can do it that why, but veeam has support for this in their product, in the UI that is already part of my work flow, this is what I consider a "OS Level or Storage level hack" it is not really part of Proxmox, but part of the underlying linux OS or Storage solution.
> Veeam use ReFS or XFS to only store one version of a block no matter how many backups are taken, this is done in the veeam software and works with their VBK file format. ZFS has some dedup which is what proxmox relies on but it works differently than Veeam, and seems to be less effective. These Block Cloning also works with the Data Format uploaded to S3 to massively save on storage costs.
> Have you ever used Veeam? I have been using Veeam since version 9.5, and hold a VMCE certification.
But sure, use Veeam, that's totally fine, and it is very well integrated quality software, but for a lot of setups PBS is way more than enough to cover their use cases.
And this thread actually started of with Proxmox VE, so just tell Veeam you want to use their software for PVE, I'm to close to this so cannot say anything, but they surely got wind of those changes coming to VMWare and surely hedge against the doom Broadcom will bring, what do you think they'd do? ;-)
Anyhow, I wish you all the best in your future migrations and hope you find a solution that fits your needs and workflows without being held hostage by your core infra provider.
[1] https://xcp-ng.org [2] https://xen-orchestra.com/#!/xo-features/webinterface [3] https://xcp-ng.org/blog/2022/10/19/migrate-from-vmware-to-xc... and https://xcp-ng.org/blog/2022/04/13/how-bedrock-streaming-mig...
I have been a paying user of WMware personally since the beginning and I have worked on their enterprise stack for several years. With the changes being made I cant see myself wanting to spend any personal money on it I and I would not recommend someone to start fresh with it, but for a lot of enterprises, they will, short term, have little choice than to pay the fees.
I have no idea about the state of Microsoft HyperV. The little I have worked with it, and that was quite a few years ago now it seemed severely lacking in the tools department.
Oracle owns VirtualBox, and even if they started to build it out (or have they?) going to Oracle to avoid high fees is not going to be a good idea.
VMware has a ton of stuff that others are nowhere near, although those offerings are useful for a minority who, as you say, will have to bite the bullet.
I really like the PowerShell applets for Hyper-V. I find them more well thought-out than PowerCLI. I do nearly all of my Hyper-V admin work at the command line.
Obviously that depends on how you are using the vmware stack. I’ve seen it being used in multiple previous employers in ways that most certainly could have been replaced by something simpler. I realize that’s because they’re not using the stack to its full potential, but that’s exactly my point. For a lot of people vmware is just “what you run vms on” because they were early in the game.
Maybe we need to adapt Open Source so that cloud providers who use it have to pay for continued development instead of just using it for free or putting a control panel on top of something and calling it their own product, but that’s the solution we came up with the first time proprietary software vendors started trying to lock the universe down into perpetual upgrade nonsense.
Perhaps it’s time to revisit Open Source licenses with a fresh, sharp pencil and an eye towards making all these systems compete with cooperative, collective-interest development again. We definitely need to find models that don’t cut the investment valuation so hard, as that’s been a huge problem for the Open Source companies.
Down markets and cost saving periods are when Open Source has seen massive adoption, as companies look to save money…
Maybe I’m wrong but it’s been on my mind a lot lately.
As someone who uses software but rarely writes anything more than automation scripts, I personally look for self-hosting and community minded software companies to have my company throw money at (cause I am high enough to influence purchasing decisions for my group).
Favorite software that a share with everyone who mentions workflow diagrams is draw.io, https://www.drawio.com/about , let's you have a local copy - format is open source - and it just works how my engineering mind wants it to work
I kinda don’t think it’d be possible to build a business like VMware’s today tbh. Certainly not after Broadcom is done with them.
I think the thing to watch here is Tanzu. In theory most of the software under that brand has an off ramp onto some other vendor. It’d be painful (ripping OpsManager out of a TAS installation is no joke) but basically straightforward technically for the right person at the right price.
Most likely people will wait for a hardware refresh to make a change, if they change at all to that will be 2 to 7 years the increased rates
and we have no idea what the increase rates will actually be, if it less than 2x most company will just eat it, that would still be cheaper than switching for many
2-4x, you will see an aggressive change in the next 12-36 mos by many customers
over 4x any you might see the aggressive pull out people are are expecting by this announcement.
Yea, a one-time cost wasn't predictable enough, instead a subscription number that will be changed each year on a whim is predictable.
Got it.
I would feel bad for them but today I was informed by my PO that the budget cuts also affect my contract so it will not be extended. The only bitter taste is that the project was meant to improve the infrastructure for citizen services. Kind of have a bad feeling for how this will turn out
1. switch to subscription model <- we are here 2. jack prices to maximize profit among the proportion of customer that can pay higher prices. 3. smaller startups migrate to other cheaper tools 4. cheaper tools slowly become mainstream 5. vmware is irrelevant and dead. ~ 7-10 years, maybe less.
The problem with this strategy here, is that unlike Photoshop (sorry gimp) there _is_ alternatives.
A big hearty fuck you to all the regulators that allowed this to go through. You are derelict and negligent in your obligations to protect the public.
I’m not sure yet where we'll end up, but the small cluster of 8 machines of ours will not run vSphere in 2025.
On the plus side, there is now a nice big clearing for new competitors to pop up. I hope open source uses this chance to catch up and offer something good to compete with nsx-t Datacenter or, at the very least, the SDN component of it. We have decent open source solutions for vms but no real open source solutions for software-defined networking or software-defined data centers. At least not any that I know of.
Either way, I think this is very good news overall. VMware was far too strong.
Fuck that. They can go and fuck themselves sideways instead. :(
Asking because although their website says its OSS it's super unclear what to download for homelab use.
If they're seriously wanting people to pay $$$ even to run it on old homelab gear, that's not going to work (for me). ;)
VMware Cloud Foundation
VMware vSphere
VMware vSAN
VMware NSX
VMware HCX
VMware Site Recovery Manager
VMware vCloud Suite
VMware Aria Suite
VMware Aria Universal
VMware Aria Automation
VMware Aria Operations
VMware Aria Operations for Logs
VMware Aria Operations for Networks
Not like open-source won, more like proprietary eats itself.
For the longest time I have been a vmware-only for baremetal. Hyper-V is ok and does the job. Virtualbox was ruined by oracle.
Oh well. Many many other virtualization platforms to choose from.
Is it time I learn Kubernetes?
I guess I won’t but it then.
Thanks.
I am so sick of this subscription game. They will lose me as a customer and anyone who will ask my suggestion in the future for virtualization. Their license upgrade model was already frustrating and now there is not even a year break from the license hell. At least I could get the license and upgrade the hosts for a few years before upgrading to the next version (ie: Vsphere 6 to 7 to 8).
It will be a pain to migrate and restructure but I won't accept subscriptions as the core of my business solution. Broadcom go to hell.
/sarcasm
> Q: Why is this good for partners?
> A: The industry has already widely embraced subscription and SaaS, and many partners in our ecosystem have already developed success practices in this area. Subscription and SaaS models provide an opportunity for partners to engage more strategically with customers and deliver higher-value services that drive customer success. It also helps accelerate their own transition to a business model focused on annual recurring revenue.
Isn't windows 11 forcing an annual prescription? That will be the real test.
Freudian slip ftw