Every VMworld conference is different, with a different tone and pace to it. At this year’s VMworld US, it felt like everything was evolutionary and very little was revolutionary. Icing that cake, VMware broke the decade-old trend of new vSphere announcements. Sure, the keynotes mentioned the next version, mostly by talking about some of the features it contains, but release dates, feature sets, and details were scarce, if available at all.
Why bring this up three months later? While VMware remains the king of data center virtualization, the public cloud—an area in which VMware is trailing—continues to add pressure. On-premises alternatives like Piston and Cloudscaling (the latter having been recently acquired by EMC), are making OpenStack a more serious contender. With VCE stagnating and the EMC Federation threatening to come apart at the seams, I’ve been thinking about what the delay of VMware’s core product means. Good sign, or bad?
Why It’s a Bad Sign
When VMware announced vSphere 5.5 at VMworld 2013, it also announced that vCloud Director 5.5 would be the last vCloud Director release the public would ever see. Yet at that time, it had nothing to replace it with. VMware vCenter is not multitenant. vCloud Automation Center was not a real replacement, either, at the time lacking credible multitenant capabilities and basic features such as the ability to import pre-existing VMs.
VMware set itself up for a form of the Osborne effect, announcing a new product too early and killing all the intermediate sales. Rookie mistake, but this isn’t a rookie. VMware has seemed a bit adrift, missing a lot of marks for its consumers. VSAN and NSX are at version 1.0, which to VMware customers means “not to be trusted in production.” VVols has been AWOL, and vCloud Suite has seen more deprecation than value-add.
Maybe VMware is coasting, like Intel did before AMD came along. We’ve seen a lot of high-level turnover; perhaps the problem is due to brain drain prompted by stock options vesting or is due to general changes in leadership. Maybe the issue simply stems from poor market research and planning. Regardless, for those seeing the dark side, missing a vSphere release is a big nail in the coffin.
Why It’s a Good Sign
vSphere is the foundation of everything built upon it. In the last two years, VMware has crammed a lot of new code into that foundation, with NSX, VSAN, VVols, the much-hated Web Client, and more. Most IT people have never been software developers, much less software developers on a large project, so they don’t understand that keeping all of that synchronized is a tremendous challenge.
On top of that, VMware is not well known for quality assurance work. It’s routine for anybody serious about running VMware software to remain with the major release that is one step prior to the current one. It’s been a real problem lately with the Heartbleed and Shellshock bugs, as anybody patching for those has seen a slew of other problems and instabilities, often unresolvable by VMware support.
Perhaps the miss of a vSphere release means VMware is finally taking such things seriously. Perhaps it’s like Intel’s tick-tock release schedule, via which it adds new features in one cycle and concentrates on code quality in the next. Perhaps VMware has bitten off more than it can chew with regard to new features, but instead of abiding by the traditional “get it out the door” mentality, it’s stopping to make sure the product is solid. Bugs aren’t just annoyances; they cost VMware money in support and engineering effort later, not to mention customer goodwill.
Personally, I hope the reason for the delay is to pay down technical debt and establish better QA. The next version of vSphere needs to support web-scale, hyperconverged clouds reliably and safely. With competitors and technologies nipping at VMware’s heels, this is a serious inflection point. If VMware can pull it off, it will entrench itself as king for the foreseeable future.