BevoMedia acquired by Vize Capital. Congrats Ryan!

Oh yeah.. and I may have scared someone out of the industry by trying to remove their eye with a small spoon.. But he was talking real shit for a long time and I told him to stop many times too. Whatever. There's this whole 'its just the internet man' kinda air of douche that goes along with it. But when you're fucking with my reputation and you're making me lose money you kinda deserve to get smacked around. Or.. a spoon to the eye. Whatever. He ended up stopping after that even though I told him to tell everyone instead.. he decided shutting the blog down and leaving altogether was safer. Shows how some people really are keyboard warriors and full of shit when it comes down to it in person and reality. Some folks from the old 202 team witnessed it along with another blogger that is no longer around (he's alive, he's just not in the industry anymore lol).
 


I should be clear in what I meant, I have nothing against self hosted except they don't scale out (which is a massive fucking problem, but eh) and I don't have anything against hosted except they have access to your data (which is a massive fucking problem, but eh).

The solution is to have built or build a tracking solution that is designed to scale and you have 100% control over (and doesn't use a language like PHP, because PHP is the devil and doesn't scale either).

How are you going to scale out the hosted solution?

Individually you're not going to scale it, but I would hope that the company running it will have the ability to do. It's the only reason to use hosted imho but even then it's not a good reason.

Why can't self hosted (in theory) do this?

It can and it's going to waste resources that would otherwise be used for something else, where it be money, time or cycles.

Kubernetes will likely make this easy (press a button) soon enough, assuming the web app is made well.

As far an I have looked at Kubernetes it would/should work, but it's still adding extra layers of infrastructure that are only being added because the self hosted solution can't scale out.
 
Acidie, it adding some sort of extra infrastructure seems unavoidable if you want to scale a system to multiple machines and have them coordinate (and survive random machines dying), right?
 
Acidie, it adding some sort of extra infrastructure seems unavoidable if you want to scale a system to multiple machines and have them coordinate (and survive random machines dying), right?

True, my use of the word infrastructure was a bit misleading.

What I should have said was, extra layers if complexity.

A lot of systems these days are virtualized out if the gate, AWS, Azure, etc and adding docker containers to the mix adds another layer of virtualization which isn't necessary and I would argue decreases to the overall stability, since the docker container is now an extra point of failure.

That and it also increases the cost of operation unnecessarily.
 
Even Microsoft has containers now ;)

I'd argue that something like kubernetes can allow a decrease in visible complexity to the programmer, due to how services are used as plumbing between components, many of which are prebuilt as containers, ready for you to use.

For example, your app can just connect to "mysql-slave" DNS name, without worrying about which machine that is, and with failed slaves automatically being disconnected from the service, therefore the app doesn't need that kind of logic built in.