Infrastructure
noun, singular.
- the basic, underlying framework or features of a system or organization.
- the fundamental facilities and systems serving a country, city, or area, as transportation and communication systems, power plants, and schools.
Bridges collapse, houses crumbles, mountains flattens and tools rusts. Something strange? No, it's called "entropy" or simply "life".
With the passage of time, things begin to deteriorate and, if they aren't maintained, eventually the will break, catastrophically sometimes. Disasters, like earthquakes or tornadoes, are not necessary, even if they help with the matter, the normal daily routine is more than enough even if it doesn't have the spectacular effects. And in this sense is even worse, because the effects are spread out on a very long time scale and as such are mostly ignored.
Sit down, wait and you'll see the body of your enemy. Unless your enemy doesn't see yours first. In both cases the problem is solved: there is no enemy anymore.
And in the same way, wait and you'll see the bridge collapse. Sooner or later. There isn't an "if" there is only a "when".
"So?" say you, "therefore" (say me), if you don't want to see the bridge coming down to meet you, you have only 2 ways: you start fixing it and keeping it up, throwing money and time at it, or you take a different road. The problem doesn't disappear anyway, it just changes.
The fact is that all those things are part of the "infrastructure", the base upon which a lot of other stuff rest to function and if this base disappear all of a sudden, everything on top of it crashes down miserably.
So of the two "solution" above, only one of the two is the good one: you need to throw money and time at it.
The problem is that throwing money at something that "works" is never perceived as a good thing. Why repainting that wall? Is good as it is. The point is that repainting also means checking it and check if there are cracks or fractures, and eventually repair those cracks before they become too depp. When it is too late to repair it, is time to rebuild it from scratch.
And of course, when you start thinking about "rebuild it from scratch", immediately other problems start showing up.
The more "basic" something is, the bigger the effect is when is taken away or changed. And when you need to fully replace something, everybody start screaming.
Ok, end of the prologue and we'll start talking about $oldandolder, those peoples had begun working LOOOOOOONG time ago, basically in another millenium (eh eh eh...) and at that time, they've decided to equip themselves with their very own mail server! That very few had something like that at that time.
But, as said before, time passes, things changes and gets older. $oldandolder chages office and take its old server along for the ride, then they change again and again and then... it is today, when the server stop working.
Crisys! Fear! Panic! Everybody run around screaming "help help" (emergency plan #1)... and at this point, the UL decide that is time to "do some maintenance" and they call us. Why us? If I got it correctly, UL was a friend of a friend of SL. So the problem ended up on my table.
Me - So, precisely what are we supposed to do?
DB - We need an estimate of the work required to bring everything in our infrastructure, possibly today.
Me - "bring" what?
DB - Well, the mail functionality.
Me - Isn't faster to simply register in O365 or Google?
DB - They also want to recover the old mails.
Me - Ok, I'm gonna ask the question, even if I already know the answer, do they have a backup?
DB - ...I don't need to answer that, do I?
Me - Not really. And since the server is dead, how are we supposed to "recover" the mail? Ask the spirit? Or phone everybody in the world and tell them to re-send their old mails?
DB - We could start by looking if the disk is still readable.
Me - Let's assume no. What's the next step?
DB - What? Why assuming no?
Me - Because is the safest assumption and we skip a lot of trouble and save time. Especially if they want to do everything for yesterday. Let's assume that the disk is gone, adios, sayonara. What's the next step?
DB - ...
Me - There is no "next" step, right?
DB - The data recovery is a required step...
Me - If it is "required" there should be a backup, if there isn't one, then the "recovery" is "nice to have but nope", so let's assume the data are gone and we go directly to next step.
DB - ...
Me - If all they want is "data recovery" why don't they go to a company specialized in data recovery? Let me guess: they want more money.
Anyhow, we go and have a look at this machine, that turns out to be a bog-standard grey-ish type tower, the type of machine that were common in 199something, with a dual hard disk (they say). With what on them? Don't know.
Me - (looking at the dust-encrusted ventilaion fan) This thing wasn't in a datacenter, right?
DB - As far as I understood has always been in an office.
Me - (looking inside) IDE! It has 2 IDE disks! Last time I saw those it was 2006 maybe! (close the box) Ok, we looked at it. Now?
DB - Hemmm... try and start it up?
Me - Didn't they tried already? Multiple times?
DB - I think so...
Me - Then we don't have to repeat the test.
DB - But...
Me - Listen, this thing has done its job, more than enough, since they never got the idea of giving it the required maintenance, because probably would've required money, they have automatically decided that what's inside isn't much important, so now it's their problem.
DB - Hammmm... What do you suggest?
Me - Let's tell them that their e-mails have flown up in the big Exchange in the sky... wait no, otherwise they think about the "cloud" and get confused. Just tell them this thing is irrimediably borked and the next time is better if they get a backup. And let them decide what to do with the mail now.
DB - They want a mail server.
Me - Ok, but a dedicated mail server? If that's the case we can spin up a virtual machine with any OS and any Postfix and we're done, otherwise they have to decide what they want. I keep repeating that a nice O365 account is better, at least they are taking the backups and we don't get to babysit them.
DB - Yeah, but it seems they wants something more...
Me - Like ?
So we go and ask this UL and it turns out that "sometimes ago" (in the Pleistocene probably), somebody had written a... thing... that read a mailbox and did... things... depending on the mails in there.
Me - Nice, so, looking at the condition of the "server", I assume you DON'T have any documentation on this "thing" or anything, as such I think the best thing to do is to start from scratch, contact this Mr Somebody and ask him if he still has a copy of that "thing" or any documentation and is willing to redo the whole thing on the new environment.
UL - We'd rather avoid to contact that person.
Me - Let me guess: you never paid him.
UL - Hmmm... let's say it was a temporary thing, more a 'test' than a real thing.
Me - Right. A test. That you kept using on production for 20 years. Well, your problem. Now is time to redo it from scratch.
UL - We wanted to recover all the data, also this "thing".
Me - "Recovering" the data doesn't imply the recovery of "everything" especially stuff that ain't data.
UL - You don't understand...
Me - Let's hear it then.
UL - That part of the system is absolutely integral in our administrative structure, without we'll lose days to get the same information and the same result!
I turn to look at the old server.
Me - If that (pointing a the machine) is the way you manage the "integral part" of your structure, I don't want to see the rest of your office.
Because infrastructure aren't eternal.
Davide
06/03/2020 11:28
