Solved How many FreeBSD installs?

Probably the closest you'll get is the figure given here. But it's very important to note that this isn't installed by default, and is thus completely voluntary.
Other than this, I don't think there is any indication of how many installations there are.
 
 
Probably the closest you'll get is the figure given here. But it's very important to note that this isn't installed by default, and is thus completely voluntary.
Other than this, I don't think there is any indication of how many installations there are.
Thanks. There are some very old releases still phoning home to report their stats. The stats gleaned from these tools are misleading because the only provide statistics from those willing to proactively provide statistics (a certain demographic).

The update and pkg servers would be another place good information could be sourced but even that would be a subset of the total (with NAT/caches/proxies and the like). This is hypothetical, as we can't look anyway.
 
To give you an idea about how much you can rely on bsdstats to have an idea about how many BSD installs there are: I just installed it on most of my systems and sent stats for the first time. For my country (Belgium) the number of systems went from 2 to 12 :)
It's a very minimal set of systems that send statistics for sure. Note: no stats were sent by running bsdstats-send manually until I added the monthly line to /etc/periodic.conf
 
The update and pkg servers would be another place good information could be sourced but even that would be a subset of the total (with caches/proxies and the like). This is hypothetical, as we can't look anyway.

I was thinking this, but as you say a large number of installations will likely cache/proxy and mask this, and also there will be a large number of installations (I suspect it'll be large) that build stuff locally from source (both FreeBSD updates, and Packages).
 
Probably the closest you'll get is the figure given here. But it's very important to note that this isn't installed by default, and is thus completely voluntary.
Other than this, I don't think there is any indication of how many installations there are.
We discussed this topic here some time ago - only very few install sysutils/bsdstats and there is no way to tell how few. I appreciate this initiative, but this is still not a good indicator.
Another problem is with the question itself - if I have a server with 4 Bhyve FreeBSD quests running, do I have 1, 4 or five installations? Not to mention that there are (say) 7 more Bhyve guests installed, but not running.
It is even hard to tell how many installations I have at home, because I have moved from one desktop system to another by replicating previous one with ZFS and putting the old drive on the shelf. This is also an installation, but idle at the moment...
 
It is interesting even if you do not think it is important. But it could be useful to, to show where the installs are concentrated and to focus on bringing particular improvements in those areas. I think detailed statistics are good when used with caution, the big issue is there is no reliable way to get the metric. If there was a default program that told how many installs there were (ethics aside) it would have to factor whether a vm counts as a real install, it tells us nothing about how often the computer is used or by who, and it tells us nothing about for how long it is installed or whether this is their first, second, or tenth installation. If I installed FreeBSD to thirty of my machines in a lab, am I single user or does that count for 30?
 
I think the better question would be - how many people are using FreeBSD? Assume that this number is much bigger than the number of users in this Forum.
I think people underestimate how much code here is used for other projects too. I recently discovered both the playstation and switch have license attributions to the FreeBSD kernel (so they use it in some capacity, whether it be as the primary kernel or as a secondary component) also include Apple and you easily have millions, if not billions of devices running FreeBSD code. It isn't even the case that these systems are just 'derivative' of, the same codebase for the JVM works on all of the BSDs AND Apple, because they are so similar and only need minor changes.
 
For the purposes of my original question, just stock FreeBSD.
We can only guess at the numbers. As has been noted already there are a number of sites that keep track of statistics but those are all voluntary, there simply aren't any "hard", reliable numbers.
 
The fact that we can't tell is something I appreciate. Many other operating systems would have an answer to this question (even if not publicly communicating that). FreeBSD doesn't phone home - at all.
This is not only a problem with the big proprietary systems like Windows & MacOS but I believe that certain Linux distributions are going this way too. Often justified as "we want to know what people use in order to improve our product(s)". Go figure.
 
The fun thing is some OSes were also considered as really, really niche - like Minix. Nobody would ever considered that as a widespread thing outside of academia.

That was until it turned out that Intel's management engine included in chipsets runs it. So the install base of Minix dwarfs the one of Linux and FreeBSD in comparison without many people being aware about it because it comes hidden and deeply buried into systems for free.
 
The fun thing is some OSes were also considered as really, really niche - like Minix. Nobody would ever considered that as a widespread thing outside of academia.

That was until it turned out that Intel's management engine included in chipsets runs it. So the install base of Minix dwarfs the one of Linux and FreeBSD in comparison without many people being aware about it because it comes hidden and deeply buried into systems for free.
Good point. Definitely the fact Intel ship so many machines means it likely has more users than stock Linux or FreeBSD but no one is really *using* Minix except Intel since that is on non exposed firmware, there are millions of L4 users too because of its usage in baseband processors. I don't find statistics reported by systems to be at all accurate, a better metric is what big users are using it, Netflix, Apple, Sony, Nintendo, are all big players using code here.
 
For perspective: Did you guys know that US census is estimated to cost $15.6 billion USD? And that's to count .3295 billion people? This works out to $47.34 USD to count just one US resident. Find someone, verify identity, whether they're alive, add to a gigantic list. Even Google would struggle to maintain a census of Android users - there's tons of people who buy cheap devices that never get updated, or specially for throwaway purposes, or they buy a device only to convert it to a different OS (Would that still count as an Android install at this rate?).
 
A lot of companies / products are also relying on some FreeBSD components while not running FreeBSD itself (at least not stock FreeBSD).

I think Jose's post is pretty much on spot here: The more interesting/relevant question should be how much support the projects gets, not how many installs there are.
This of course depends on what your goal of the question is.
 
I think folks are so used to applications and operating systems with telemetry running all of the time that they forget that not all software developers think it's a great idea to build it into their products. Thankfully, my FreeBSD instances are not phoning home unless I ask them to. Good luck getting a number that's even close to accurate.
 
Back
Top