[hitpointdrew]

Like what? There is no real open source comparable option to VMware if are running your own hypervisors.

KVM? Not even close, you need a host of other tools to get what you have with esxi and vcenter. You’ll spend months just finding and configuring a disjointed hodgepodge of tools to get what you could have setup in a day. And even then you have a fargile system that is likely to break any time you update one tool in your stack.

[sofixa]

> There is no real open source comparable option to VMware if are running your own hypervisors

Well the first question is, do you actually need hypervisors? In a lot of cases, the answer is no, or not for many of the workloads. In others it's 100% yes.

Then, after you've decided you actually do need a hypervisor, there's actually tons of choice - Proxmox (very good and advanced KVM wrapper), Nutanix's AHV, oVirt (future kind of up in the air), OpenStack, KubeVirt, XCP-ng.

Problem is, many VMware users are set in their ways and want an exact and 1:1 replacement, without even considering they were only doings things that way because that was the only tool they knew and had at their disposal, not because it's actually a good way of doing things. Virtual Machines are just a means to an end, and a clunky one at that. VMware are actively pushing you away, time to start paying attention and considering what the organisation's actual needs are, and how are they best served. (And unless you're doing VDI, or almost exclusively using third party appliances delivered as VM images, that's not virtual machines).

[fred_is_fred]

>Problem is, many VMware users are set in their ways and want an exact and 1:1 replacement, without even considering they were only doings things that way because that was the only tool they knew and had at their disposal, not because it's actually a good way of doing things.

Correct. You have entire teams in your average Fortune 500 who have built the entire career on being the VMware team.

[sofixa]

Well, unfortunately for them, it's the time to evolve or get left behind while others in the organisation (either users who are unhappy or higher ups looking at budgets) try to push for change, with good reason.

[icedchai]

KVM works great. Do you think major cloud providers are building everything on VMware?

[gertrunde]

KVM does work great, but that doesn't invalidate the rest of the comment.

Do you think major cloud providers haven't engineered the whole "You’ll spend months just finding and configuring a disjointed hodgepodge of tools to get what you could have setup in a day." step?

[icedchai]

Of course, they build their own tooling. They'd have to build their own to operate at scale. For a small business or home lab operation that needs dozens of VMs, that already exists (as pointed out by another poster, who mentioned libvirt and virt-manager.)

[gertrunde]

libvirt, virt-manager, et al are all perfectly suitable tools for an average member of the HN audience, or someone playing with a home lab, or any number of other technically proficient users - but those people are not the people who are buying vmware's products.

And virt-manager comes across from their docs as a bit on the basic side, proxmox and ovirt look a bit more polished - but again, companies are often wary of running their entire infrastructure on something where there isn't someone they can blame if it goes wrong - it's more about managing the risk than anything else.

I would absolutely love it if these sorts of companies used and relied on open source technologies more, but it's unlikely that they would pay for the staff with the relevant skills to manage it all, or pay enough to retain those staff that get trained up internally. (And I'm somewhat hopeful that this broadcom/vmware mess will cause more resources to flow into open source projects, or spark new developments in this space - cloud isn't always the answer to everything).

[icedchai]

You are right. I don't understand why companies will continue to shell out big $$$ for VMWare licenses and training, but not invest the time to learn open source. Maybe start with a POC running a few VMs on KVM/libvirt. It doesn't have to be all or nothing.

It also doesn't help that many more traditional "IT" folks I've met don't want to learn anything new. I still know guys that don't understand IPv6, for example.

[hitpointdrew]

>KVM works great. Do you think major cloud providers are building everything on VMware?

Major cloud providers have huge teams developing their compute as a service, that is what they do, that is the bread and butter.

I don't have the resources and team size to hack our own hypervisor/on-prem cloud solution.

I have a team of 3 people, that is it. I can't put 5 people on designing and maintaining a hodgepodge solution of various tools that get us to where vmware already has us. I need a turn-key solution that a small team can manage.

[bdhcuidbebe]

Theres clouds built with KVM, you dont know your stuff.

[hitpointdrew]

> Theres clouds built with KVM

No shit, they have huge teams and this is their bread and butter. I have 3 people. I do know my stuff, which is why I would never use KVM and try to match that to the features VMWare gives me with a team of 3 (including myself).

[Teever]

Well yeah, this is at least as much a side effect of people not using open source solutions in the first place as it is the reason that people don't use open source VM solutions.

If more people had used open source solutions in the first place then they wouldn't be in this situation that they find themselves in now.

[vishesh92]

I am running Apache CloudStack with KVM in my homelab. Took me a couple of hours to set it up.

It might not have all of the features of VMWare but it just works.

[samsk]

Proxmox is quite good, it's not VMWare, but hey, it's not VMWare ;-)

[athorax]

OpenStack is pretty okay these days