Personally, the choice of Linux distribution is only related to the distribution of the design concept and their own compatibility, and other learning ah what the relationship is not big. The same kernel, most things are similar, no difference
- Ubuntu: Desktop application-based gnu/linux operating system; it means that I exist because we exist; Ubuntu is based on the Debian release and GNOME desktop environments, unlike Debian, which releases a new version every 6 months. A long-term support version of LTS is released every 2 years; Development concept: Ubuntu program emphasizes ease of use and internationalization so that it can be used by as many people as possible. (from wiki)
- CentOS Redhat to be outside of Rhel, the other reason to support a fedora is to get some new technology/software into the Fedora community, and so the new software has been tested by the community, after the iteration, to ensure the stability before entering into Rhel
Because the old version of the software does not have some new features, no special fancy effect, but after years of practical testing, is actually available.
And, in addition to the Fedora test site, Redhat also has a large number of QA staff to do a lot of testing for each Rhel release (which will eventually go to CentOS), and a hardware compatibility team to specifically guarantee the hardware compatibility of each RHEL, which are rhel/ CentOS as a production system of the operating system must, all of this is to pursue the production system of stability and reliability.
Redhat Internal is the division of the team into project and product two parts
Project team is dedicated to working in the open source community and doing their own
The product team is then tested based on the release of a project (functional test, performance test, compatibility test, etc.), backport fix, documentation, and finally as a product to paid customers
For example, Hibernate Community Edition released 4.0, but the product of hibernate does not follow, will stay in 3.5
4.0 of the Community edition after a period of user feedback (this feedback is very efficient, because the community used a lot of people, the community Jira basically a lot of jira by open, including bugs, incompatibilities and so on), and then release 4.0.1, then feedback, and then release 4.0.2, Re-feedback, re-release 4.0.3
At this point, after a few iterations of community feedback plus bug fix release, basically 4.0 of this big version of the problem will be discovered, then the product team began to intervene, they will have a special QA to do more detailed testing, and then according to different needs, See is can only do bug fix backport or need to put some user needs of new feature also port to the old 3.5, then publish a 3.5.1 product
Can see, a product of the hardships of the birth process .....
So, if you have the manpower and resources to do enough testing of the new version of the software, and have the ability to fix it yourself (assuming it's all open source software), then there's no problem in pursuing the new version (but don't say cost savings, which is a lot of cost), If you do not have this ability, or honestly use the results of others (this paragraph excerpts from strong Liu's answer http://www.zhihu.com/question/29191794)
Linux distribution version
