In software release cycles, version numbering has become a de-facto standard: major releases come out with a integer number increment, while minor releases have a decimal increase. Apart from few exceptions (Microsoft using years lately or Linux kernels) the vast majority of software vendors and even the open sourced ones follow this standard. In this way, users can easily understand if the new build is a minor upgrade, usually with bug fixes and few enhancements, or a major one, and act accordingly.
In the last years, VMware has used version numbering for its main product in a completely different way. The last two upgrades from .0 to .1 has been major releases; thinking about new functionality or big changes, 4.1 has changed vCenter from 32 to 64 bits, and lately 5.1 brought SSO and the Web Client among the most visible changes (not considering “under the hood” ones).
I got many customers complaining about this behaviour. They thought it was a minor release so they installed those upgrades and then had to deal with the changes. As usual, a RTFM warning is needed: even if it would have been a minor upgrade, even a single patch, customers are not excused for not have read the release notes. None the less, this non-standard approach to version numbering from VMware seems to create confusion on customers.
Basically, you can find a pattern in VMware releases: every new version is a major release. Think about it: 3.5 – 4.0 – 4.1 – 5.0 – 5.1. All of them brought important changes, and can be considered major releases regardless of their numbering. Fixes and minor upgrades only come with patches and they are numbered as uX (U as Upgrade), like 4.1u2 for example.
But is this something understandable at first sight by not well informed customers? We are used to read blogs, forums, attend conferences and webinars. Do all customers do the same? Or they just read the newsletters coming from VMware or an email from their consultants?
What you think? Again, it’s up to the final user to ALWAYS read release notes. But can VMware nonetheless improve this numbering schema?