This week, VMware finally GAs the latest and greatest version of its flagship product, vSphere. We have now reached the lofty heights of version 6.5. It has the usual improvements. The vCSA can now handle updates natively, has high availability, and runs on PhotonOS. Virtual machines can be encrypted.
Now, I do not intend to deep dive into all the new features; you can read the What’s New document as well as I can. That said, with this release, I do not have that buzz I used to get with a new vSphere release. The reason, I feel, is that although the new features are welcome and extend the capability of the platform, they most likely will not be widely employed. On the whole, they will be utilized for niche use cases. vSphere is no longer the crowd puller it used to be. Like an aging rock star who is still trying to fill stadiums, it just seems a little sad.
The hypervisor is now passé, with regard to vSphere; it has met the vast majority of users’ needs since version 5.0. The newer features are really just sprinkles on your ice cream. With the release of Server 2016, Hyper-V is now good enough, and RHEL-V is, too. XenServer, if Citrix can get its marketing and sales teams into gear, is also a viable product. I cannot find myself getting excited about the hypervisor any more.
VMware’s core technologies—vSphere, VSAN, and NSX—are only interesting together and are only complete when together, which brings up the next issue: they are just too expensive. vSphere costs $2,875 per CPU for Enterprise Plus; VSAN is $2,875 per CPU for the advanced version (the only one that is really worth having), and NSX Enterprise is $6,995 per CPU—this means more than $15.5 over and above the cost of a dual-CPU server. You can purchase VMware Cloud Foundation; the cost is unknown, as it is limited to VxRail, VxRack, or prequalified VSAN-ready nodes and networking switches. For more information, see the VMware Cloud Foundation product page. This is all well and good, but at the end of the day, vSphere, VSAN, and NSX do not make a cloud.
Although it is a core and interesting technology, even VSAN is just another storage target, another hyperconverged infrastructure play, a point solution to deliver local distributed storage. Like all VMware products, it denies the plethora of nonvirtualized x86 devices and non-Intel devices that are still out in the wild. That said, it is moving in the correct direction with its new ability to be used as a native iSCSI target for other storage consumers.
That NSX needs to be integrated deeper into the product is clear, but it also needs to be expanded out of the virtual arena. It needs a physical presence. Yes, this is “available” with the use of third-party partners like Juniper and Arista Networks, but these are just gateway devices to convert NSX into legacy networking protocols. What is needed is the ability to install it on dumb devices in the same manner as an OpenSwitch.
So, is this just another VMware-bashing article about how the writing is on the wall for the company? No; think back to my earlier post here, where I outlined my views regarding one of VMware’s potential future paths. VMware’s future is in the overlay tools, vRA, Horizon View, vCD, and management stacks. Historically, its messaging has been fragmented and incoherent. However, I believe that VMworld Barcelona was a nexus point with regard to this issue.
For VMware to succeed, it needs to continue driving its core products into more places, but those costs are now proving problematic. It is now used to having more bells and whistles than the other players, but they are not being used. It is sad to say, but hypervisors are now being chosen on price, not technology, and VMware cannot win that war. To win the new battle—the one for multi- and intra-cloud management—there is no need to control the hypervisor. VMware just needs to realize that.