"Just do X" is advice that is almost always less than useful.
IT departments that use Horizon with instant clones and have several compute clusters with AutoHA, DRS and NZX, a sprinkle of vSAN etc. etc. etc. will take a week just to translate all of what they need to do to Proxmox terms - and then a month to figure out if Proxmox or XCP-NG can even do all that.
But if the 10x price increase is actually typically, they might actually look into it. Will still take years to migrate.
As much as I love Proxmox, it’s not a VMware replacement unless you’re a small shop with knowledgeable people. There is a huge gap between “homelan where I can play with ZFS” to “enterprise-grade workload management and redundancy” that it only partially addresses.
Maybe in a few years. I know that if I was working there, I would be trying to close as much of that gap as possible while the migration opportunities are still around.
VMware is to virtualization as windows is to computing.
Essentially this is broadcom turning a cash cow into a milk truck for a handful of large megacorps that cannot evolve past a dead product and will pay handsomely for licenses without question. No one but VMware administrators legitimately consider esx for virtualization, instead the market is cloud providers or onprem kvm/cockpit/proxmox/lvs/anything else open source that doesn't bleed quarterly earnings dry for a chance to spin up a SQL server.
VMware is _everywhere_. Every big company you can think of has a significant VMware estate somewhere. This is going to accelerate as companies move back on-premise.
And for good reason. ESX is by far the most performant hypervisor out there and it's dead easy to use.
But being 95% performant compared to native speed is not worth a lot if the "only 85% performant" alternative costs millions less, since you can often spend half of those millions on more or better hw and still come out with a decent profit.
It's cool and great to have the best product, but if the "only" edge is that if you spend a million bucks on license, you save 10s or 100s of K in hw expenses, the business proposition is actually not that great.
I get that the whole ESX ecosystem has a lot going for it, but as some of the VM loads are getting more and more dumbed down, being the one with most features and some more percent better perf is not going to win out in the end, as I see it.
That 15% gap matters a lot for the kinds of companies that Broadcom is targeting (and many others that they are not), and they are okay with spending millions to get it.
Also, you get support with those millions. That's a requirement for these kinds of companies as well.
That said, I agree that there are many shops that will migrate to something like Ryzen, throw KVM on that and call it a day (or do what I did and use Graviton spot instances for anything stateful and small Graviton Lambdas for the rest). Those suckers are fast.
And that's before we touch software-defined networking a la NSX and virtualized storage a la vSAN. NSX, ESX and vSAN work together to provide near-line-rate networking and networked storage performance with a really nicely packaged frontend. Yes, you can use paravirt-io and ceph and get nearly there, but it's "nearly" and setting this up has a significantly higher skill bar for sysadmins. That alone is a dealbreaker for small shops that aren't tech forward.
I'm really curious if there are some good academic studies on the incentives for execs to take on large, "marriage of behemoths" tech acquisitions. Given that so often, perhaps a majority of the time, the end result is just a destruction of value, I'm curious as to why these are entered into so frequently. Perhaps it's a case of "90% of them are destructive but the 10% winners are huge winners"?
In the case of Broadcom they actively seek technologies that are extremely difficult to pull out of large organizations. Then they make their money back by gutting all sales staff and r&d and jacking up prices for everything. They believe they can make their money back and then some off the Fortune 500 and they’ve bet correctly almost every time.
Horrible way of doing business, not sure why it’s allowed, but it works. Just barely better than corporate raiders.
Getting back a 69$ billion price tag that way seems optimistic, but maybe I'm just pessimistic about the cash to be gained by being entrenched in large corporations.
Either way it does not feel like this strategy will open new lines of revenue for them.
>Getting back a 69$ billion price tag that way seems optimistic, but maybe I'm just pessimistic about the cash to be gained by being entrenched in large corporations.
VMware had $13B in revenue in 2023. I expect the average customer's cost to double over the next 24 months - so put that at $26B doing nothing beyond increasing costs (and forcing customers into a subscription with products they don't need or want). Then cut all sales staff, R&D, and gut support. They will have paid off their investment in under 5 years. That's not even including the sale of end-user compute which is predicted to go for $4b.
>Either way it does not feel like this strategy will open new lines of revenue for them.
I guess that depends on what you mean by "opening up new lines of revenue". It is already opening up a subscription line of revenue they have not had previously - they've dumped all perpetual vmware licenses and are forcing everyone to subscriptions.
If you mean taking the VMware product and adding new features to increase the revenue stream of the existing customer base via an expanded product: of course not. They aren't interested in improving the product, they are interested in milking every last penny out of it at the lowest possible overhead. Why do you think the ink was barely dry on the deal and they already killed ESXi free edition, and all perpetual licenses?
Execs do mergers simply because stock price go brrrrr, which is what their bonuses are about. When the projected gains will fail to materialize and the acquisition mismanagement will be laid bare, they'll be somewhere else anyway.
Occasionally it's because the acquired company is really on its last leg, and they're trying to cash in by leaving some sucker to hold the bag. Also occasionally, because the acquiring company has been given a bunch of money and they have to show that they're serious about growth, so they'll just buy anything and anyone even tangentially related to their business.
These are the sympathetic explanations. The unsympathetic ones involve security fraud and anticompetitive behavior; but, of course, those would be wrong for anyone to allege. No honest exec would ever accept under-the-table payments, or a promise of future windfall or lucrative job, to greenlight a takeover by a rival; we all know execs are honourable people.
Most likely licensed at exorbitant prices or killed, Broadcom likely can’t get a sizable chunk of money for it given the current state of desktop virtualization, and they are all about tossing small customers to the ditch.
The only IT people I know complaining about this VMWare stuff are the same people who have proudly talked about how “you can get by without programming” well sorry folks but those days are over, and the sysadmin who can adapt their company logic to run on AWS Spot Instances is keeping his job, while the rest of the infra team is getting canned because the pretty gui didn’t tell them how to do their job.
When I was doing these migrations for side work I was shocked at how basic most of these VMware shops were. Some terraform and a couple of bash scripts is usually enough to lay off the entire VMware team (3-10 people). Made absolutely no sense the amount of money some of these companies were spending and how little work the infra guys were doing for like, 100 servers.
This one time I got a call from the owner wondering when we’d start billing, turns out their AWS bill was so cheap it didn’t even hit finance that it was their new infrastructure cost, and the dude was just in shock at how incompetent his old team was, he was surrounded by “the cloud is a scam” people for years so he never thought that they were just full of shit.
What an aggressively myopic, narrow view of the situation. The reason "this VMware stuff" is a huge issue is because Broadcom are just absolutely fucking up relationships between every part of the VMWare ecosystem: the vendor relationship, the support relationship, the licensing system, the trust in a product reliability roadmap, just everything. This creates chaos in any org relying on them, at any size and efficiency. Chaos costs a fortune. The reasons that any org has for keeping their server operations on-prem and reliant on VMware are not as blithely dismissible as you are indicating. In fact, the reasons are frankly irrelevant anyway, as even the most impeccably managed migration plans to whichever big 3 cloud vendor you'd like are going to be thrown at least somewhat askew by the recent actions from Broadcom. The volume of clear, unambiguous discussion and media coverage that this is getting goes a long way to smooth over the pain of mitigation strategies with nontechnical leadership.
Broadcom hasn’t screwed up anything with the big contracts, only SMEs. That’s the entire business plan. Did you even read the article? This has been going on for a while.
Nothing I’ve said is wrong, either. It’s almost entirely affecting the small shops who don’t have resources. Big buys are renewing contracts and calling it a day because VMware caters to them.
Oh boy, are you mistaken. Big contracts with upcoming renewals have already been approached and are being hit with 3x-5x increases on hypervisor licensing alone (source: I used to be in the ISP/hosting business and have recently consulted on an infra provider acquisition that is hung on this).
I know plenty of people in the sales pipeline handling VMWare renewals, they are getting 0 pushback from these massive increases because of my above comment of teams not knowing how to use anything else.
That's just because those customers have already decided to move but will take the renewal for 1 year until they re-skill. Many admins are looking to learn other tech.
26 comments
[ 5.3 ms ] story [ 66.3 ms ] threadIT departments that use Horizon with instant clones and have several compute clusters with AutoHA, DRS and NZX, a sprinkle of vSAN etc. etc. etc. will take a week just to translate all of what they need to do to Proxmox terms - and then a month to figure out if Proxmox or XCP-NG can even do all that.
But if the 10x price increase is actually typically, they might actually look into it. Will still take years to migrate.
Maybe in a few years. I know that if I was working there, I would be trying to close as much of that gap as possible while the migration opportunities are still around.
Essentially this is broadcom turning a cash cow into a milk truck for a handful of large megacorps that cannot evolve past a dead product and will pay handsomely for licenses without question. No one but VMware administrators legitimately consider esx for virtualization, instead the market is cloud providers or onprem kvm/cockpit/proxmox/lvs/anything else open source that doesn't bleed quarterly earnings dry for a chance to spin up a SQL server.
And for good reason. ESX is by far the most performant hypervisor out there and it's dead easy to use.
It's cool and great to have the best product, but if the "only" edge is that if you spend a million bucks on license, you save 10s or 100s of K in hw expenses, the business proposition is actually not that great.
I get that the whole ESX ecosystem has a lot going for it, but as some of the VM loads are getting more and more dumbed down, being the one with most features and some more percent better perf is not going to win out in the end, as I see it.
Also, you get support with those millions. That's a requirement for these kinds of companies as well.
That said, I agree that there are many shops that will migrate to something like Ryzen, throw KVM on that and call it a day (or do what I did and use Graviton spot instances for anything stateful and small Graviton Lambdas for the rest). Those suckers are fast.
And that's before we touch software-defined networking a la NSX and virtualized storage a la vSAN. NSX, ESX and vSAN work together to provide near-line-rate networking and networked storage performance with a really nicely packaged frontend. Yes, you can use paravirt-io and ceph and get nearly there, but it's "nearly" and setting this up has a significantly higher skill bar for sysadmins. That alone is a dealbreaker for small shops that aren't tech forward.
Horrible way of doing business, not sure why it’s allowed, but it works. Just barely better than corporate raiders.
Either way it does not feel like this strategy will open new lines of revenue for them.
VMware had $13B in revenue in 2023. I expect the average customer's cost to double over the next 24 months - so put that at $26B doing nothing beyond increasing costs (and forcing customers into a subscription with products they don't need or want). Then cut all sales staff, R&D, and gut support. They will have paid off their investment in under 5 years. That's not even including the sale of end-user compute which is predicted to go for $4b.
>Either way it does not feel like this strategy will open new lines of revenue for them.
I guess that depends on what you mean by "opening up new lines of revenue". It is already opening up a subscription line of revenue they have not had previously - they've dumped all perpetual vmware licenses and are forcing everyone to subscriptions.
If you mean taking the VMware product and adding new features to increase the revenue stream of the existing customer base via an expanded product: of course not. They aren't interested in improving the product, they are interested in milking every last penny out of it at the lowest possible overhead. Why do you think the ink was barely dry on the deal and they already killed ESXi free edition, and all perpetual licenses?
All the folks at the top typically win on their options/etc, they don't care, they got paid.
Execs do mergers simply because stock price go brrrrr, which is what their bonuses are about. When the projected gains will fail to materialize and the acquisition mismanagement will be laid bare, they'll be somewhere else anyway.
Occasionally it's because the acquired company is really on its last leg, and they're trying to cash in by leaving some sucker to hold the bag. Also occasionally, because the acquiring company has been given a bunch of money and they have to show that they're serious about growth, so they'll just buy anything and anyone even tangentially related to their business.
These are the sympathetic explanations. The unsympathetic ones involve security fraud and anticompetitive behavior; but, of course, those would be wrong for anyone to allege. No honest exec would ever accept under-the-table payments, or a promise of future windfall or lucrative job, to greenlight a takeover by a rival; we all know execs are honourable people.
When I was doing these migrations for side work I was shocked at how basic most of these VMware shops were. Some terraform and a couple of bash scripts is usually enough to lay off the entire VMware team (3-10 people). Made absolutely no sense the amount of money some of these companies were spending and how little work the infra guys were doing for like, 100 servers.
This one time I got a call from the owner wondering when we’d start billing, turns out their AWS bill was so cheap it didn’t even hit finance that it was their new infrastructure cost, and the dude was just in shock at how incompetent his old team was, he was surrounded by “the cloud is a scam” people for years so he never thought that they were just full of shit.
Nothing I’ve said is wrong, either. It’s almost entirely affecting the small shops who don’t have resources. Big buys are renewing contracts and calling it a day because VMware caters to them.