If is not broken...
If is not broken, don't fix it. I am pretty sure everybody has heard this sentence at least one (million) time.
Usually, this is used when somebody, in a fit of activity, decide to put his mittens on something, breaking it completely (with collateral damages). Of course hindsight is 100% accurate.
But if we go to look at things from really close, we (when I say "we" I mean "we humans") are constantly fixing stuff that ain't broken.
Because what is NOT broken right now is going to be broken pretty soon.
Entropy, is the term used by scientists and whomever pretend to be "smart", everybody else simply calls it "time passes". And it effects everything that exists. Sure, NOW is not broken, but if you look close, you can see the cracks starting to form.
Why do you think it's called system "maintenance" ? Because systems require constant maintenance. In many circumstances is small stuff, but slowly things gets more and more corrupted, until they actually crack, that is, they stop working altogether. Of course there is always the possibility that something crashes down hard, because of internal or external intervention, in these cases the term is "incidents". But even if it doesn't go down with a bang, there is always the possibility that we decide to replace it before the it does. Is normally in these circumstances that the aforementioned sentence is mostly used and it loses its "wiseness".
Because, yes, we are FIXING something that is not broken (yet). To avoid that it does break.
At this point, usually, everybody start screaming not only that "you don't fix yada yada", but there will be "stuff to change". That is true. It doesn't matter if the purpose of the "new" thing is the same as the old one, you can bet your left ball that something will be different.
Let's look at software for example.
Now, software usually is either broken from the beginning or it doesn't break on its own (let's forget the problems connected with peoples that click on everything without looking, like the "delete" button). Experts could object that software is ALWAYS broken from the beginning, because software always has bugs, small or ginormous. And software normally tends to have some "problems" that the more you use it the more is annoying, like that bump on the chair that doesn't look that big in the beginning, but after 3 hours it starts to be a real PITA.
And all softwares, especially the "serious" ones, every now and then needs to be updated, to apply patches and fixes for the aforementione bugs or... other stuff. And we could discuss on the fact that the real purpose of most "updates" is that whoever produce the software would like to keep eating, that can be understood.
But in some circumstances, when things changes drastially, you can have problems.
Let's say, for example, that a beautiful bridge, that was beautiful when it was new, but now it ain't, was in need of a lot of maintenance, and anyway is now too small for the volume of traffic it should carry. Well, you could decide that is time for a new bridge. And here we can clearly say that we are fixing something that is not broken, but it could be broken very soon, with drammatic consequences, especially if you happen to be on the bridge when it goes from "not broken" to "broken". But re-make the bridge is going to be a problem. Firstly, cost, of course, but secondly, if you decide to take down the bridge to rebuild it, in the mean time, there will be no bridge.
Of course you could leave the bridge there and build a new one (not-so) nearby, but this also has problems, you need to build another piece of road, with added costs and you also need to consider whatever is around. If you "upgrade" the bridge so it can carry more traffic, what about the rest of the road?
And after this "roady" introduction, we are back talking about $screamandyell of whom I spoke sometimes ago.
After the Great Catastrophe, of which I told above, the "old" system had been quickly re-instated by UL, and then... and then, apparently, nothing happened anymore for a while. Besides the fact that, as "punishment" for the failed "secret" operation, UL had "taken over" some maintenance activities on the old system.
When I say that he had "taken over", I mean, of course, that he had organized with $screamandyell that they would send the requests to us (that means ME of course) and we (plural Majestatis) would have done them.
However, since those "maintenance activities" turned out to be lots and lots of MANUAL changes on the messages and the archives and they were not that "sporadic" but quite frequents, after having a look at that stuff I let UL know that 1. No, and 2. No. and 3. Seriously what the fuck were you thinking?
UL - But I've already told SL that we would have taken care of these things!
Me - Then it's better if you get cranking, because I am not going to do this shit.
UL - But you're the sysadmin!
Me - Correct. I am a sysadmin, and this has nothing to do with 'sysadminning'. This is not system administration or maintenance, this thing is beyond "maintenance", this thing is at "take it behind the shed and shoot it" stage. And there is already a brand new system that is perfectly working and it only needs a good testing to be put into production.
UL - He... last time it didn't looked like it was working so well...
Me - That because you ignored TESTING because you didn't wanted to talk with the customers.
UL - But if we tell them that there will be changes they could decide to leave!
Me - I think that's hard to believe, borderline bullshit.
UL - What?
Me - Those peoples, the customers I mean, are using the system as a data transmission mean between firms, as such every change require some testing and cannot be done so easily, as we discovered. Now, even if they decide to leave right now, they'll have to either rebuild the system on their own, because it's an integral part of their business, go look for another one or rebuild their entire business to not use such thing. The latter probably require a lot of time and effort and money, so I don't think is doable, the rest, still require them to change part of their infrastructure to cope with a new system. So it doesn't really matter what they decide to do, they will have to apply some changes. At this point, if they decide to "leave" is not because we decided to change, there is something else.
UL - But...
Me - And the fact that there are all those changes in the messages and archives make me think that maybe, the old system isn't really "ideal" for whoever is using it. Maybe it WAS perfect in the beginning and now is no longer so. So we could start asking those people how they actually use and then alter the new one to better suit their need. Maybe the correct thing to do is not to make the new one "as close as possible" to the old one but "as close as possible" to what is needed.
UL - ...but...
Me - This also means that we'll probably need to re-implement part of the new environment and test it. With the cooperation of the customers.
UL - ...but if it works why do we want to fix it?
Me - ...because the concept of "it's working" is a lot subjective. As you should have understood by now.
Davide
24/02/2021 09:33
