There's no shame in VMware quitting the public cloud
Cisco and HPE couldn't make it work and Virtzilla probably dodged a server refresh cycle
Analysis VMware quitting the public cloud by selling vCloud Air to OVH looks like failure, but is the best possible sort of failure because the company still has excellent prospects to turn a quid from the cloud, without having to operate one.
Which is not to say that VMware is out of the woods. It clearly thought it could run a public cloud and that turns out not to be the case. That leaves it with a plan to keep building out a software-defined infrastructure stack it feels will deliver more value the more of it you implement and the deeper you go into hybrid cloud. That plan now rests on some muddled product lines that I'll get to later.
In the short term, however, the vCloud Air's failure won't be too painful. The service probably made VMware a few tens of million dollars a year, not chump change but also not a hard-to-replace chunk of VMware's US$7bn total. vCloud Air also had the potential to cost VMware a lot of money: it was launched in 2013 and therefore was almost certainly getting close to end-of-life for the first set of servers and associated infrastructure. If selling to OVH means VMware has dodged that responsibility and associated capital expenditure, the financial impact of the deal could be pleasant.
VMware might cop some heat from customers that relied on vCloud Air for exotic cloud workloads: The Register's virtualization desk has been told that VMware did some tricky stuff like running DOS workloads in vCloud Air, and that customers liked the fact it would hand-hold them as they got that up and running. There's nothing we see in this deal to stop VMware helping OVH to keep that stuff alive.
Another positive is that VMware has freed itself from conflict between its own cloud and the 4,000-odd vCloud Air Network partners who run vSphere-powered clouds. Those partners may lament the fact VMware won't have skin in the game and therefore stops market-making, but there will still be lots of noise for them to latch onto once VMware-on-AWS fires up later this year.
Gartner research director Michael Warrilow also pointed out to El Reg that OVH's European footprint is a plus for European businesses, especially in the time of Trump.
Consider, also, that VMware is in good company as a cloud flop, given HPE and Cisco both created and then canned clouds. Both have deeper pockets than VMware by quite a distance too. If those two titans couldn't make a public cloud public viable, VMware's not in bad company.
But let's get back VMware's product muddle. Most of the vCloud Air Network today runs more-or-less vanilla vSphere. But VMware also has a Cloud Foundation bundle of vSphere, Virtual SAN and NSX that it says is the best way to build a hybrid cloud (and is also a fine way to get VMware users to buy in to VSAN and NSX, thereby extending VMware's on-premises footprint). Gartner's Michael Warrilow also pointed out to El Reg that VMware's Photon Controller is basically a re-write of vCenter for hyperscale operations, which could also have potential for service providers.
Just how Cloud Foundation, vSphere-on-AWS and Photon Controller fit together is currently a little mysterious. All offer a chance to make money from clouds. The three also overlap and muddy the waters, which VMware can do without.
At a lower level, the deal means VMware has some explaining to do to those who signed up for vCloud Air assuming they'd get one throat to choke. Such customers' lives just became rather more complex, no matter the assurances offered by OVH and VMware.
Speaking of complex, OVH also has a tricky job ahead of it to make the service work, because much of vCloud Air resides in third-party data centres from the likes of Equinix and Telstra. It must be galling for a cloud provider like OVH to have inherited obligations to rivals. Unpicking those arrangements without disruption is an unenviable task because flubs or outright failures have the potential to hurt customers.
But it's hard to see much else in this deal that will immediately be bad for customers. Which means even though the mere fact of the deal is an admission of vCloud Air's failure, its one that won't mean pain for VMware and its customers. ®