VMWare has a proven history. They have set the bar and when potential problems are brought to their attention, they address it. Everybody else is simply a VMWare wannabe.

Bonus points as with a google "domain" - all I do is point my MX records at google and I get gmail on the backend.

You can only have 10 users for free, but you essentially have an unlimited number of 'groups' - and when you set those groups so that "anyone on the internet can post" they turn into forwarders.

Meaning you have 10 discrete accounts on the domain, but more aliases than you'll ever need.

First of all let me say VMware makes one rock solid hypervisor. Out of all of the hypervisors I have used theirs is the least prone to issues.

VMware has a couple of really BIG problems in their platform.

1. Their management tools are windows centric and so is Virtual Center for that matter
2. Their licensing model is confusing as hell and requires a spreadsheet to figure out what you need without overpaying
3. They have so many products that it gets downright confusing to determine which one works for your purpose.
4. They use "old school" sales tactics that just don't work for more modern companies.

When I am engineering a solution and have a problem to solve I am presented with many challenges to present VMware as a solution. Finding the product that suits our needs, Figuring out what license would suit our needs, getting a quote from the vendor without a lot of harassment after the fact trying to close the sale. Rather than deal with all of that many people have found that the open source projects like KVM and XEN are good enough for their needs. Not to mention the huge numbers of free cloud products such as Openstack that gives you enterprise features "for free". At the end of the day I don't care what product get's used as long as the problem is solved with the minimal amount of budget and effort.

A small startup does not want to deal with legacy software and maintaining licensing and dealing with windows boxes. They do great with the "enterprisey" douchebags with their complex setups that cause more outages than they solve but lean and small companies don't want their stuff.

The reality here is the world is slowly changing. Big monolitic companies are failing because their business models are unsustainable without cheating and people are getting fed up with the cheating. VMware has to answer a question to themselves. Do I want to serve the needs of the dying dinosaur companies or do I want to be in business in 10 years?

Two years ago, VMware was the only serious player in the enterprise hypervisor market. It could demand a price premium and had weight with other software platform vendors to demand support.

Now, with Hyper-V being somewhat more mature and with the Xen product line, VMware is falling into a competitive market. Naturally, there will be an erosion of market share in that case.

The bigger threat that faces VMware is the same threat that faced Netscape in the 1990s-- VMware is a competitor of Hyper-V which Microsoft can include "for free" in its server operating systems. And Microsoft still has the same monopoly influence over the major hardware vendors (to discourage pre-installs or reseller agreements). And it can control the licensing for its operating systems to inconvenience VMware customers (you have to buy a separate license for each ESX VM, but if you run on Hyper-V you get 10 VM licenses for free) and/or control its support of its enterprise application stack (We'll only support Exchange/Sharepoint/SQL Server/Link/IIS/whatever if it's running on Hyper-V. If it's ESX, please reproduce the problem on physical hardware to make sure it's not an ESX issue)

--Joe

Sounds like a pretty sweet setup.

So, when a production server refuses to boot after you've just done a P2V migration, who do you call for support?

^ This.

If you don't need HA and can live with a little guest downtime to migrate from a downed host to a cold spare, and have the wherewithal to figure out how to use the host efficiently, ESXi costs nothing and shared iSCSI or NFS storage that performs reasonably well is pretty cheap.

I've seen such a setup: a bunch of commodity Xeon 5520-equippped boxes attached to an EqualLogic SAN stuffed with SATA drives. It could have been cheaper if Equallogic would support customer-provided disks, because the premium on rotating rust, while not at EMC's level, was still pretty steep.

+1. Million.

EMC/VMware got greedy with the licensing for v5 and anyone who used some the nicer features (vRAM oversubscription) could potentially find themselves paying a lot.

The drastic limit applied to vRAM in the lower tier editions of ESXi 5 smacks of greed. They might not get caught on this now, but it was bald-faced enough to make people think about going to v5 and to evaluate HyperV instead. I might add that HyperV looks a lot more attractive since MS gives away Windows guests licenses that you'd otherwise pay for with VMware. If you virtualize a lot of Windows servers, going to HyperV could save you huge dollars, whereas vSphere 5 will cost you more.

Consider NetWare: It was a better directory/file/print server than NT, but Microsoft made a compelling argument that NT4 and, eventually, Win2K were good enough that it was worth losing some features, especially if it meant cutting down on the number of platforms and boxes to manage. The pitch for HyperV is very, very similar.

HyperV might not be so hot now, but VMware can't get complacent. Now is the time to put the boots to HyperV, not cede market to it.

I don't know if you're a marginal case or not, but I had a VMware employee that they felt that they had to raise prices because server core and memory densities were getting to the point where they were going to start losing revenue.

They said they had seen a tendency among very large customers to actually cut CPU licensing. Earlier adopters had started smaller on older hardware that didn't have the massive amount of CPU and RAM that's common today and the old per-CPU licensing model meant that growth in these environments meant more licenses.

But with memory and CPU densities growing, these customers now need fewer licenses, even though they still have VM growth, because they have been buying 24 core, 128GB boxes to replace 16/64 or smaller boxes. Each new box can replace 2 or sometimes 3 older ones.

IMHO, they're just making up for the general reduction in price they had once they started offering EssentialsPlus ESXi. This was much cheaper than ESX 3.5 with vCenter and vMotion licensing.

I also wonder if maybe they shouldn't have switched from per-socket to per-core licensing, but charging less per core than they had per socket. This would have allowed stable revenue with increased core counts.