Our normal use case is running on linux (unless you’re in the UK part of my company in which case you vastly prefer running on windows and need to lean on the US aka me and my boss, to help you install your product on linux)
I’m no stranger to the advantages of encapsulation and I acknowledge they have their uses.
In this specific instance though, the problem they’re solving is that like 4 people don’t wanna learn linux, but the market demands they install on linux.
That’s it.
(also those 4 people seem wholly unable to retain anyone they hire that understands linux)
Yeah, if the sole reason for using containers is to avoid having an admin who knows Linux, that is BS. Docker containers are useful, but I wouldn’t trust anyone to use them properly who doesn’t at least have a basic understanding of what Linux stuff they actually do under the covers.
Ya know funny enough, I can see a way the docker WOULD actually help us.
I don’t know the fancy term for it, but when traffic spikes and we want to spin a bunch of new things up to cope, docker would be a lot faster than our ansible deployment process. Thing is though, to actually implement that you gotta actually write the software that detects the spike and sets to spin up the new instances.
That’s the exact thing we’re not doing right now in favor of just getting it running smoothly in docker.
(I maintain it runs smooth as… something smooth when installed on linux)
That is definitely part of the reason we are (were?) moving to a provider that already has it. Point is, though, I’ve been told using encapsulation actually helps if you wanna build it on your own.
Docker may still have some benefits if running multiple Docker instances in a single VM comes out to be cheaper than running multiple VMs while still handling the load you toss at it. A lot of this does come down to what provider you use, your particular load expectations, your budget, and so on.
We use docker + kubernetes where I work to keep out public api’s up during the christmas rush. We had a few years where we basically spent every october building and deploying machines to handle the excess load. This annoyed me enough in 2017 that I just dockerized it and threw it into kubernetes and then forgot about it. (I am oversimplifying it a bit, but not much) the christmas season came and went and I checked on it once just to make sure that kubernetes was contracting when it was supposed to.
My main issue with it is that it works well enough that I don’t need to look at it much more than once every few months. This lead me to basically forgetting nearly everything I did to set it up or do any debug to it. So when something does go wrong I find myself digging through documentation going “where the hell is the config for this thing?!?!”
edit: we also run on GCP, and chose it specifically because of their kubernetes facilities.
What is not laid out in the article, since Ed couldn’t firmly make an evidenced connection - the announcement of orders being open for the 35K Model 3 came out just before Tesla’s almost billion dollar convertible debt to the government came due, a debt which many were wondering if they could actually pay off, or even make a good-faith payment on. Curious, that.
Like. The closest I get to preordaining is buying something sight unseen.
Like when FromSoft released their new game, I bought it like the day after it came out with like no reviews and before the community reached a general consensus over how good it is.
This represents the best case scenario for me. Buying something quickly. Not before it exists.
True. As far as I can recall, the only Kickstarter I’ve ever contributed to was for the Pride Pets pins, and that was only because I knew the person behind it semi-personally. I’ve looked at other Kickstarters with curiosity, but I can’t remember ever contributing to any of them.