MTBF and lifespan are not the same figure and cannot really compared. MTBF is normally used for repairable failures of devices and is a statistical figure only. When you have 200 devices operating for 1000 hours and one of them fails, there is an MTBF of 200.000 hours. But that in no way indicates what would happen after 200.000 hours (or 100.000 or whatever).
Unfortunately the MTBF is usually specified without additional information of an expected lifespan so users often think that the given MTBF is a useful indication of expected lifespan (considering that after 200.000 hours the unit is expected to have failed). But that is wrong.
Of course we are talking about specification, not about personal experience. Factory says MBTF is expected to be 200.000 hours. That does not say that lifetime will be 200.000 hours! It says that after some specific operation time (without that extra info the MTBF is basically useless) a certain number of failures can be expected. With only the number 200.000 you do not know if it is one in 200.000 after 1 hour, or one in 200 after 1000 hours, or one in 20 after 10.000 hours.
But what you know for sure it does not mean "they will fail only after 200.000 hours".
But we do have a single data point on real life of the devices.
The first devices are almost 10 years old.
Which means at the most 87.600 hours.
Since there are not tens or hundreds of posts on the forum of people shouting about their RB3011 having died, I have to presume that the percentages I drew out of my hat in my earlier post might be conservative.
I postulated that:
5% of devices will die in the 0-15000 hours range <- what I call infant mortality
another 10% will die in the 15001-87.600 hours range <- temporally up to 10 years of life
So, after 10 years of life, 85% should still be operational, my guess is that if this percentage was lower than this, we would have seen many more reports on the forum than what we have.
If you buy an used one now, it is already a survivor.
another 15% will die in the 87.600-127.900 hours range <- the next 5 years or so
another 30% will die in the 127.901-199.900 hours range -<- the following 9 years or so
Hopefully some 40% will slowly die afterwards.
But the points #4 and #5 are irrelevant if you are only trying to understand if spending 75 ā¬ today (or anticipating 15ā¬/year for 5 years) is a good deal or not.
More or less, what I am hypothizing is that if you buy a new device you have 15% probabilities that it will die in 10 years time.
While if you buy a ten years old device you have roughly the same probability that it will die in half the time (5 years).
But if you buy an used RB3011 now, very likely it is more recent than 10 years or however it has less than the max 87.600 hours, and you should be fine for the next 5 years.
My experience with old MikroTik devices (and with electronics in general) is that what fails first is the powersupply.
Especially those plugpacks that are coming with routers without integrated powersupply usually fail. Fortunately the routers are very tolerant towards supply voltages and any other plugpack with DC center-positive in the 10-30 volts range will normally work.
Which isn't actually a router failure, since it's a pretty generic, easily replaceable, external device. New wart, and instantly back up (not even a migration). If you have multiple devices, keeping a spare wart isn't a bad idea either, esp. for core devices, no matter what the brand. (Had a full spare, pre-configured Cisco backup when I ran Cisco as well . . . . $h1t happens! Best to be prepared!)
Yes, this is also my experience with PC's, over the years what fails is usually the PSU.
I am writing this post from an old PC (running XP, SP2, and yes, I know) that was bought in 2008, so roughly 18 years old that - for some reasons - is always on (though not really used interactively as workstation anymore), so surely it takes mains fluctiations, over and under voltages, spikes, etc..
Over these years I changed on it some 3 ATX PSU's, last one in january 2026.
I would also go a little bit farther and say that - at least in my environment - there isn't an overly marked difference in durability between "known brand" power supplies and el-cheapo (though not exactly "no-brand") ones.
I now consider changing these PSU's every five/six years or so like "ordinary maintenance", and I always have a spare one.
The other things that (much less often, let's say every ten years or so) tend to have issues are fans (even if I do have a once a year scheduled maintenance of cleaning them), this might be due of course to the dust in the environment and mileage may greatly vary.
When (IF) I can, I tend to prefer passive cooled thingies.
For the same reasons, In the case of many Mikrotik or non Mikrotik routers/switches with their external adapters/bricks I don't consider a "failure" (even if it is one strictly speaking) the need to replace one of them every few years.
It is much more aggraviating when a "custom" PSU fails (internal) I had a lot of troubles over the years with non-standard PSU's like the ones that are sometimes used in thin clients (nowadays many have also an external brick) or in small cases (like mini-itx ones), over the years I had to actually repair a few of them since a spare with the right dimensions and power were not anymore available (and - at least for me - finding the problematic component(s) and replacing it/them has been a nightmare, each time).
It's much easier to get proper certificates for such devices as they do not have 110/230V "inside" and therefore use only safe voltages what implies that users can safely open them themselves according to law
Yup! No need for UL (or whatever) for the base device, and can either cert a wart for multiple units, or offer one from another vendor (rebadged) that is already approved.
(I kinda hate warts, but they do serve thier purpose.)
It's not the same as the company's support.
Some MT devices are supported by OpenWRT but it does not mean that they have endless support.
OpenWRT also has some limits on chips and memory supported.
I would say that it's a "sparkling water or still water" type discussion . We should be happy that MT (manufacturer) supports many old almost unusable devices. It means that somehow it limits itself in evolutionare or revolutionary (that is bad) changes but it give us, the users, the ability to test/use new functions with already owned gears (what is good).
