> > In large corporate/enterprise environments with hundreds or thousands > > workstations and laptops a centralized management tool is a must and > > any home-made scripting solution hacked by those two geeks working in > > the basement is too fragile solution so something proven and tested > > would be needed. > All this stuff, whether corporate or Linux or even budding new OS's are > written by > Geeks. Where they work, and how much they get paid and by whom has very > little impact > on value. Some practical examples. Early Fedora and Red Hat Enterprise Products are based around the up2date. That was written internally for enterprise purposes by Red Hat. Fedora took the decision to also take on yum, which was written externally (initially for Yellow Dog) and then much updated. It wasn't many Fedora releases before yum went from "grudgingly added by community demand" to "default", and shortly after to "also used as the base package update technology for enterprise products" Some of the largest networks and clusters in the world are managed by custom, in-house developed solutions. Practically it is usually about two things, neither of which are about who wrote the code - Having proper process You can write it in the basement but you still need to document it and write it to an appropriate standard, design, security policy and with agreement from the correct people. It has to produce the correct logs and meet the documented goals. - Doing proper testing Goes without saying I would hope. (and in some cases) - Having a third party for support/blame/testing/.. The latter basically being an outsourcing decision to gain economies of scale, ideally by using the same tools as others or at least harvesting the benefits of a common pool of knowledge and expertise you don't have to own internally. Alan
