jb_fvwm2 said:
I can imagine that paragraph posted in a Linux forum resulting in several users and maybe even some datacenter managers proceeding to google "this vs that" -- and as a result moving their machines over to FreeBSD... the only thing stopping that now being the idea of an alternative, as in "FreeBSD? it has a forum even? "
I'm assuming you're referring to my paragraph. But perhaps you are not familiar how enterprises work. I'm pretty sure nobody is going to "move over their machines to FreeBSD" just because they read something in a forum, or blog post, unless their business does not depend on a stable and well understood environment. They might "fire up FreeBSD alongside their existing ecosystem" to see what is it all about. That's what we're doing.
We are a small company, delivering some SaaS solutions and managed hosting, few dozen servers, nothing big. We are running FreeBSD for a year now alongside existing Debian, Ubuntu, CentOS and Gentoo installations because we're testing and trying it out. We are certainly not going to base any decision without extensive testing and strategic planning. It's not just installing and using the system. We have tools, configuration templates and systems, monitoring, failover and redundancy, all tailored for the systems we use, all requiring change of some kind in order to "switch". I can imagine much bigger companies having even more resistance, testing requirements and above all paperwork, to change.
Above all, our testing for our use cases is telling us we can't switch.
Linux is faster for us because of PostgreSQL regression on FreeBSD 10. We're a PostgreSQL shop. Linux is more secure for us because we run well documented and understood SELinux and AppArmor, and the Linux kernel has had for years some security features (like ASLR) that are yet to come in FreeBSD (10.1), and MAC/RBAC on FreeBSD is completely unknown to us and pretty much undocumented. Linux as a distro with supported non-base software is more stable for us because it is stabilized by many, many people before release and there is much less work for us to test and deploy updates as opposed to FreeBSD ports which are moving as fast as a rolling Linux distro would, which requires far more testing, more manpower, build servers, staging/testing servers, etc... We're a Python shop and our software is grown on Debian 7 with Python 2.7.3, while on FreeBSD the 2.x branch is 2.7.8. Now, believe it or not, we can't switch without changing the code because some bugs were fixed and some features introduced between those two versions that is breaking our code on FreeBSD. That, of course, is
absolutely not FreeBSD's fault, I'm just trying to show that there are many unforeseen problems and "switching over" is not just a matter of firing up a new VPS... Which in itself would also be a problem because our hosting company is not supporting FreeBSD. And if you've been in the web hosting industry as a consumer of hosting services as long as we have, you'd know that once you settle with a hosting company that works for you, finding another (that might support FreeBSD) is a nightmare that also requires many months of testing under various circumstances to see how they fare.
We can use Debian 7, Ubuntu 14.04 LTS, CentOS 6.x for years to come without having to violate our systems with systemd, and I'm pretty sure our clients would like us NOT to change anything in that department without significant amount of objective reason. We will continue to use FreeBSD in parallel and test it, perhaps there are use cases and scenarios which we haven't yet encountered that would significantly swing the opinion over for the switch.
Don't get the above as something against FreeBSD. Quite the contrary, despite all that, we continue to use FreeBSD in testing. I'm just trying to paint the picture how "switching over to FreeBSD" is far easier said than done. I'm sure the bigger the enterprise, the bigger the inertia toward any kind of change. So no, I don't see any mass exodus happening over systemd to anything - another Linux or BSDs.