Striving to become a master Go/Cloud developer; Father ๐จโ๐งโ๐ฆ; ๐ค/((Full Stack Web|Unity3D) + Developer)/g; Science supporter ๐ฉโ๐ฌ; https://coder.today
When the kindergarten goes to visit a museum or play in the park, you need a couple of grown ups to take care of you.
The grownups are the ones that
decide where you go today (on which cluster/server provider)
send and filter messages to you (decide what outside messages to deliver to you or not)
tell you where to sit (decide on which machine to run)
make up pairs of children to play together (locality rules, you may want to group 3 python instances with 1 mysql replica, so they could talk to each other in an optimal way)
keep an eye on you (health checks)
One big difference between this example and an orchestrator, is that k8s doesn't hesitate to "kill" instances that misbehave and start new ones.
Docker runs containers. Kubernetes orchestrates containers. Orchestration is how you stitch multiple containers together into a greater whole: for example, you can have several application containers (and scale out to add more application containers at runtime) talking to a database container. It's possible to do this with Docker alone, but prohibitively complicated at the low level of individual containers. The language of "pods", "services", "deployments", and so on that Kubernetes employs lets you operate at the level of multiple interacting containers instead.
We're a place where coders share, stay up-to-date and grow their careers.
Top comments (4)
When the kindergarten goes to visit a museum or play in the park, you need a couple of grown ups to take care of you.
The grownups are the ones that
One big difference between this example and an orchestrator, is that k8s doesn't hesitate to "kill" instances that misbehave and start new ones.
!remindme 3 days
Wait, this isn't Reddit.
Kubernetes is just a Theme Park :)
Blogged a longer explanation here
My favorite material about this:
Docker runs containers. Kubernetes orchestrates containers. Orchestration is how you stitch multiple containers together into a greater whole: for example, you can have several application containers (and scale out to add more application containers at runtime) talking to a database container. It's possible to do this with Docker alone, but prohibitively complicated at the low level of individual containers. The language of "pods", "services", "deployments", and so on that Kubernetes employs lets you operate at the level of multiple interacting containers instead.