Title: Obsolete in the IT crossfire | |
Author: Solène | |
Date: 09 July 2021 | |
Tags: life linux unix openbsd | |
Description: | |
# Preamble | |
This is not an article about some tech but more me sharing feelings | |
about my job, my passion and IT. I've met a Linux system at first in | |
the early 2000 and I didn't really understand what this was, I've | |
learned it the hard way by wiping Windows on the family computer (which | |
was quite an issue) and since that time I got a passion with computers. | |
I made a lot of mistakes that made me progress and learn more, and the | |
more I was learning, the more I saw the amount of knowledge I was | |
missing. | |
Anyway, I finally got a decent skill level if I could say, but I | |
started early and so my skill is related to all of that early Linux | |
ecosystem. Tools are evolving, Linux is morphing into something | |
different a bit more every year, practices are evolving with the | |
"Cloud". I feel lost. | |
# Within the crossfire | |
I've met many people along my ride in open source and I think we can | |
distinguish two schools (of course I know it's not that black and | |
white): the people (like me) who enjoy the traditional ecosystem and | |
the other group that is from the Cloud era. It is quite easy to bash | |
the opposite group and I feel sad when I assist at such dispute. | |
I can't tell which group is right and which is wrong, there is | |
certainly good and bad in both. While I like to understand and control | |
how my system work, the other group will just care about the produced | |
service and not the underlying layers. Nowadays, you want your service | |
uptime to have as much nine as you can afford (99.999999) at the cost | |
of having complex setup with automatic respawning services on failure, | |
automatic routing within VMs and stuff like that. This is not | |
necessarily something that I enjoy, I think a good service should have | |
a good foundation and restarting the whole system upon failure seems | |
wrong, although I can't deny it's effective for the availability. | |
I know how a package manager work but the other group will certainly | |
prefer to have a tool that will hide all of the package manager | |
complexity to get the job done. Tell ansible to pop a new virtual | |
machine on Amazon using Terraform with a full nginx-php-mysql stack | |
installed is the new way to manage servers. It seems a sane option | |
because it gets the job done, but still, I can't find myself in there, | |
where is the fun? I can't get the fun out of this. You can install | |
the system and the services without ever see the installer of the OS | |
you are deploying, this is amazing and insane at the same time. | |
I feel lost in this new era, I used to manage dozens of system (most | |
bare-metal, without virtualization), I knew each of them that I bought | |
and installed myself, I knew which process should be running and their | |
usual CPU/memory usage, I got some acquaintance with all my systems. I | |
was not only the system administrator, I was the IT gardener. I was | |
working all the time to get the most out of our servers, optimizing | |
network transfers, memory usage, backups scripts. Nowadays you just | |
pop a larger VM if you need more resources and backups are just | |
snapshots of the whole virtual disk, their lives are ephemeral and | |
anonymous. | |
# To the future | |
I would like to understand better that other group, get more confident | |
with their tools and logic but at the same time I feel some aversion | |
toward doing so because I feel I'm renouncing to what I like, what I | |
want, what made me who I am now. I suppose the group I belong too will | |
slowly fade away to give room to the new era, I want to be prepared to | |
join that new era but at the same time I don't want to abandon the | |
people of my own group by accelerating the process. | |
I'm a bit lost in this crossfire. Should a resistance organize against | |
this? I don't know, I wouldn't see the point. The way we do computing | |
is very young, we are looking for a way. Humanity has been making | |
building for thousands and years and yet we still improve the way we | |
build houses, bridges and roads, I guess that the IT industry is | |
following the same process but as usual with computers, at an insane | |
rate that humans can barely follow. | |
# Next | |
Please share with me by email or mastodon or even IRC if you feel | |
something similar or if you got past that issue, I would be really | |
interested to speak about this topic with other people. | |
# Readers reactions | |
ew.srht.site reply | |
# After thoughts (UPDATE post publication) | |
I got many many readers giving me their thoughts about this article and | |
I'm really thankful for this. | |
Now I think it's important to realize that when you want to deploy | |
systems at scale, you need to automate all your infrastructure and then | |
you lose that feeling with your servers. However, it's still possible | |
to have fun because we need tooling, proper tooling that works and | |
bring a huge benefit. We are still very young in regards to automation | |
and lot of improvements can be done. | |
We will still need all those gardeners enjoying their small area of | |
computer because all the cloud services rely on their work to create | |
duplicated system in quantity that you can rely on. They are making | |
the first most important bricks required to build the "Cloud", without | |
them you wouldn't have a working Alpine/CentOS/FreeBSD/etc... to deploy | |
automatically. | |
Both can coexist, both should know better each other because they will | |
have to live together to continue the fantastic computer journey, | |
however the first group will certainly be in a small number compared to | |
the other. | |
So, not everything is lost! The Cloud industry can be avoided by | |
self-hosting at home or in associative datacenter/colocations but it's | |
still possible to enjoy some parts of the great shift without giving up | |
all we believe in. A certain balance can be found, I'm quite sure of | |
it. |