Fault Tolerance and Fargo
Fault tolerance in VMware has been pretty useless up until now. If you honestly believed that you needed the weedy, single vCPU version of fault tolerance that existed in previous versions, you were either in a very bad place or unaware of the alternatives. But oh, fault tolerance that supports four vCPUs is an absolute game changer. I will go so far as to say that it is industry redefining.
4 vCPUs makes fault tolerance actually useful. It also means that VMware has cracked multiple CPU fault tolerance, and that the difference between 4 and 16 is only a matter of the hardware and interconnects that we have available to make it all go.
There is now no excuse not to be 100 per cent virtualised. Even for applications upon which lives depend. It's time for companies with those sorts of workloads to be talking to VMware, their vendors and everyone else in the chain to see if this solution is right for them... and all of us who depend on them.
The catch – and there's always a catch – is that 4vCPU fault tolerance will massacre your network. 10GbE might be the official bare minimum quoted by many, but if you are seriously looking at this tech don't touch it unless you have at least 40GbE.
Meanwhile "Project Fargo" is finally out in the open. This is basically the concept of the "golden master" from the VDI side of the fence brought over to servers. If it's been a while since you've played with the full suite of what's possible with modern VDI tech, take a look. It's actually really cool. If, however, you're a hard-boiled VDI nerd, you've been screaming "when is all this stuff coming to server workloads" since the beforetime anyways.
VMware in 2015
VMware's new focus is on building hybrid cloud capabilities into every aspect of its offerings. Everyone else seems convinced that businesses can be convinced to cough up outrageous sums of money for cloud subscriptions to and VMware isn't going to be left out of the party. VMware has signed a deal with Google to ensure that they have the cloud capabilities to go toe-to-toe with Azure and AWS, incidentally giving Google it's only real shot at having a hybrid cloud answer to those same competitors.
VMware is rumoured to be moving towards smaller, more regular releases (at least for some products) and will be dogfooding those products in their hybrid cloud service before releasing them to the general public. Changes in licensing models are always being discussed, but the results of the ROBO edition experiment are not yet certain.
What's crystal clear from virtually all my sources at VMware is that the pressure is mounting to drive up average revenue per customer, even while customers are chafing against existing costs. Expect VMware to pound the vCloud Air and NSX drums very loudly during 2015. Right behind them will be VSAN and vRealize.
The hypervisor is a commodity, and VMware knows it, but VMware doesn't want to be pigeonholed as "just selling management tools". VMware want you to know that they are ready not simply to be a part of your datacenter. They are ready to be the entire datacenter.
VMware have the hypervisor. What everyone else has is nice, but if we're being honest, nobody else even comes close. VMware has the management tools that don't suck. Hell, even the web client is better than System Center, not that this is hard.
VMware has NSX, and NSX is good; they are absolutely ready to handle your most complex networking needs...even if those needs include layer 2 extention into vCloud Air. And thanks to vCloud Air and NSX, VMware currently has the only big name hybrid cloud solution that doesn't suck. Though they had better watch out, as eager competitors like Yottabyte are building strong challengers.
VMware has VSAN, and VSAN is also good; VMware can provide enterprise storage for your datacenter themselves, or through their army of partners attaching through NFS 4.1, VVOLs, through kernel integration or as virtual hyperconverged offerings. Though here again VMware needs to be cautious; competitors like Nutanix, SimpliVity, Maxta, Scale Computing, Yottabyte, NimBOXX, HP, Gridstore, Atlantis, and many (many!) others are seeking a slice of the pie.
Many of these hyperconverged competitors offer not only solutions that work on VMware's platform, but also on KVM, Hyper-V and there are even some talking about Xen. Many of these competitors offer features VMware currently does not. Those who are working in the KVM space are a special concern, as many of them are integrating with the kernel, removing VMware's loudest marketing option from play. They are also legitimising that platform by putting real money into making management interfaces that don't suck, and bringing hyperconvergence to Openstack.
This then brings us to VMware's collision with Openstack. Openstack is all about giving service providers the ability to build their own cloud. VMware is (mostly) offering this ability too. But VMware also has its own cloud interests to protect, and it would really like it if everyone would (pretty please) dump their VMs into VMware's hands...along with all that nice subscription money.
How is this all going to play out? Nobody knows; that is as much a function of international politics as anything else. Powerful forces – mostly notably the Americans and the Brits – are hell bent on stripping us of every last vestige of privacy. Companies (and individuals) who think about such things for more than a few seconds tend to have some problems with that. Especially since – and let's be honest here – nobody can trust the Americans not to engage in industrial espionage.
Microsoft is busy trying to strangle its own channel, jacking up prices for service providers and partners and making the only hybrid cloud in a can offering they have inflexible and insanely priced. HP is busy tearing itself apart, Oracle is crafted from the fundamental evil at the center of the universe, both Amazon and Google don't care, and IBM couldn't find the future with two hands, a Sherpa and a GPS.
That leaves the open source community via Openstack, VMware and Dell to provide cloud software and services to the 6.5bn+ people who aren't American or British. Mirantis, Cisco (via Metacloud) and Piston Cloud – just for starters – will show you that Openstack is ready to meet this challenge. Dell's plans are a complete mystery (and good on 'em for that,) which leaves us with VMware.
vSphere 6.0 is unrepentantly badass. vCloud Air is coming along nicely, and all of the other pieces of the puzzle are evolving steadily as well. This is great and wonderful, but the question that will hang over VMware for all of 2015 is how they will handle the hybrid cloud.
Will VMware go the Microsoft route and pay lip service to their channel and partners while silently working to kill them off? Or will VMware be both combatant and arms dealer, keeping prices low enough for service providers to build competitively priced VMware-based clouds?
To whom will "the other 6.5 billion" belong? With technology like VMware has demonstrated in vSphere 6.0, they are VMware's customers to lose. ®