VMware's past holds the key to the future of Microsoft's Azure Stack
Virtzilla's mistakes show how Microsoft must not package its hyperconverged cloud
By most accounts VMware's first stab at hyperconvergence, the EVO:RAIL software-defined Nutanix clone, was a decent set of software.
But it failed, quickly, and has been quietly removed from the frontline.
The reason why is simple: in its first incarnation VMware insisted that EVO:RAIL buyers acquire new vSphere licences. Which was a big mistake because it turned out that most people who fancied EVO:RAIL were already vSphere users who liked the idea of adopting hyperconverged infrastructure. Insisting that EVO:RAIL meant the acquisition of new vSphere licences therefore made no sense.
VMware eventually made vSphere licences portable but in the months between launch and that change fatally annoyed plenty of potential buyers, crippling the product along the way.
Which brings us to Azure Stack, the on-premises cloud-in-a-box Microsoft last week delayed until 2017. With VMware's errors in mind, The Register asked Microsoft if users will be able to bring existing Windows licences to Azure Stack.
Redmond politely told us it doesn't have an answer. “We’ll have more to share about share licensing and pricing details at a later date,” a spokesperson said.
Let's hope that promised “more” includes portability between current Windows Server licences and Azure Stack. Let's also hope that Azure Stack is a two-way street and that there are no barriers to moving workloads or data from cloudy Azure to on-premises Azure Stacks. It makes no sense to restrict movement, but as VMware's mess and Microsoft's hardball licensing history both show, vendors can get sometimes get greedy. ®