I'm not surprised at the Pivotal acquisition. VMware is determined to succeed at Kubernetes. There is already a lot of integration with Pivotal's Kubernetes distribution both at a technical as well as a business level.
We found that Pivotal's kubernetes makes it easy to run kubernetes on top of existing vmware infrastructure, so our ops team could continue to manage their normal vmware stack, with an admin interface to manage kubernetes clusters that live within esxi. They could provision us devs with a cluster for a specific use, and hand us the keys to do the rest.
It was pretty cool, except for the fact that we discovered kernel incompatibility with the underlying esxi version/os that brought down the entire deployment repeatedly a few days before the event where we intended to stress test the product.
The use case is definitely there for larger orgs who want to share hardware, but you also get the "on prem cloud" quick wins. Our experiment obviously failed in a low risk way, but i think the idea is sound.
It’s a framework that makes it hard for one side to accidentally depend on something they shouldn’t.
Here i am just referring to the ease of using iaas offerings like s3 buckets but through on prem hardware. Devs build against drop in blob storage apis as if we lived in aws, and my ops guys still manage the backing vms.
"Like a cloud but closer to you" :)
Private Cloud buildout plays into Dell's interests, but if their execution is worse than even one Public Cloud provider, then they risk this becoming an exit strategy.
You can make a lot of money off of being an exit strategy, and for quite some time. But when the evacuation is over you have nothing.
Honestly if Dell does this well enough I could see them using some of IBM's old tricks. One in particular I'm thinking of is that IBM would overprovision the customer (installs are more expensive than dormant hardware). Fast forward to your next crunch time and you're oversubscribed, you call up IBM and they send you a piece of code to unlock the rest of the servers. No scheduling, no deliveries.
If I'm running my own server room, it turns out having a bunch of hardware around I don't use is prohibitively expensive (in part because corporations use funny math and account for labor costs in weird ways). Which kinda bums me out because I used to get excited about ideas like running servers where it's cold to save huge amounts on electricity.
But if I'm leasing hardware from Dell, then the math is simpler. Labor and travel time are the same color of money as the hardware. Installing a whole rack of servers and only leasing you a half or quarter rack at a time (and using some of the extras for hardware failure) might be more cost effective. And I'm sure IBM knew what they were doing. The immediacy of picking up a phone and having 'more hardware' today versus scheduling it three weeks out (and having the team decide to prioritize optimization work) must be more lucrative. If a little bit like drug dealing...