Le 21/01/2021 ? 23:18, Valeri Galtsev a ?crit?:> No, I already streamlined Debian routine installation (workstations and number > crunchers), and servers run FreeBSD since loooong ago, so I'm all set, and much > better than in the past ;-) > > Thanks though.Debian has an average of two years[*] per support. Oracle has ten like upstream RHEL. Choice is pretty clear to me. [*] one year after subsequent release, so an average of one to three years depending on installation date -- Microlinux - Solutions informatiques durables 7, place de l'?glise - 30730 Montpezat Site : https://www.microlinux.fr Blog : https://blog.microlinux.fr Mail : info at microlinux.fr T?l. : 04 66 63 10 32 Mob. : 06 51 80 12 12
On 1/21/21 4:50 PM, Nicolas Kovacs wrote:> Le 21/01/2021 ? 23:18, Valeri Galtsev a ?crit?: >> No, I already streamlined Debian routine installation (workstations and number >> crunchers), and servers run FreeBSD since loooong ago, so I'm all set, and much >> better than in the past ;-) >> >> Thanks though. > > Debian has an average of two years[*] per support.There is "stretch", which is equivalent of more known as LTS of Ubuntu: 5 years. And then, there is easy in place upgrade from regular or "stretch" to next release. But no, I will not argue against uniqueness of 10 year life cycle of RedHat. I just said that my life [with Debian] will be no bigger hassle than it was [with CentOS]. The only difference of Debian is: it has vast collection of everything, so you really need to make your own choices. But if it's done once, you can in one go tell next installation to install all the same software (packages). Of course, I, being a simple guy, had much simpler life with CentOS, just choose all software groups that sound relevant... (jus grossly exaggerating ;-) But with huge collection like Debian one (or like FreeBSD ports are, or macports for MacOS) once you spent time shaping system to your preference, you are done, and all next systems are rather routine, almost as unattended as RedHat/CentOS kickstart install is. Valeri> > Oracle has ten like upstream RHEL. > > Choice is pretty clear to me. > > [*] one year after subsequent release, so an average of one to three years > depending on installation date >-- ++++++++++++++++++++++++++++++++++++++++ Valeri Galtsev Sr System Administrator Department of Astronomy and Astrophysics Kavli Institute for Cosmological Physics University of Chicago Phone: 773-702-4247 ++++++++++++++++++++++++++++++++++++++++
On 1/21/21 5:50 PM, Nicolas Kovacs wrote:> Debian has an average of two years[*] per support. Oracle has ten like > upstream RHEL. Choice is pretty clear to me. [*] one year after > subsequent release, so an average of one to three years depending on > installation dateSo, I want to address the "ten years of support" albatross.? On the surface, ten years of support sounds like a big win; it certainly did to me back when it was first introduced.? I have found that the reality is far more nuanced than that.? I have found in my own career that the "ten years of support" argument has made me lazy in keeping up with newer technologies and methodologies, stagnant in my own server and workstation deployments, and increasingly frustrated once the five-to-seven year point has passed in what I can't do or can't build because "ten years support!? Stability! Stability! Stability at all costs!" For my uses and purposes, Fedora's six month cycle is too fast (I've been on that roller coaster before, no desire to go back to it). CentOS Stream's continuous release cycle is too fast, especially in the kernel ABI department.? I believe that, for my uses at least, a two-to-five year cycle is going to be the sweet spot.? And the fact of the matter is that CentOS and the ten-year cycle isn't nearly as stable as you might first think; install CentOS 7.0 on a test VM and carefully compare to 7.9, especially on the workstation side with Firefox and Thunderbird! Further, when it's budget time, updating stagnating services running on a stagnant OS becomes an easy mark for cutting from the budget, because "ten years!" - until those ten years are over and you find out that you've just delayed all the effort into one lump instead of spreading it out a little bit each year or two (or three to five). But ten-year stagn^H^H^H^Hupport also makes me less marketable if I were to need to change jobs, especially if that ten-year stability has calloused my learning skills to the point that I feel personally threatened by major changes to, say, the init system underneath everything. So, in my career, I'm not sure relying on ten-year support has been a good thing.? YMMV as I'm sure there are places where ten years of support really is critical; just not for me.