Docker Containers on PodSpace

We automatically create a virtual private network for every project that you create on PodSpace. All containers within a project are connected to the project’s virtual private network, and thus they are completely isolated from other projects and from the outside world.

So how will other people access your application, you might wonder. With just a couple of clicks, you can define and expose virtual load balancers which direct traffic to your containers. Based on what you have defined, our load balancing infrastructure will route incoming requests from your users to your application.

Speaking of load balancing: should you want to run multiple copies of a container, you can do that with a single click too. This is how it works: when you deploy a container to PodSpace, it starts the container in a Pod. Here is how it looks like after you have deployed an application:

All you have to do is to click on one of those two arrows you see at the right side of the deployment status box, and PodSpace automatically starts up new copies of your Pods if you have increased, or terminates them in case you lowered the number.

The virtual load balancers that you have defined for that application (BTW, they are called Services if internal, or Routes if exposed to the outside world) automatically adapt to the new topology.

But you don’t have to stand there and stare at your resource usages so that you can click on the up arrow when your application suddenly gets a burst of load: you can set up an autoscaler so that PodSpace scales your pods up and down as needed. Easy.

By defining health checks, PodSpace can determine if your application is working properly. If it finds that an instance is not working, it is instantly put out of service (that is, no user requests get forwarded to it from the load balancer), and restarted. After restarting, the health checks are run again and the container gets put back into service.

You can define health checks in simple terms such as "consider my app healthy if it responds with HTTP 200 to this URL", or "consider my app healthy if it is listening on port 8080". Or you can define advanced heuristics by telling PodSpace to call a shell script inside your container to determine application health.

Investigating problems within Docker containers can be notoriously clumbersome without the right tools. With PodSpace, however, it is easy: you can open a shell session into any running container with just a click in your browser.

Let’s imagine that your application is in production, serving requests continuously. Let’s say it has to serve so much traffic that it’s scaled to multiple copies. Now what happens if you have made a new version and you want to deploy it to production?

We provide automated processes that handle the version rollouts so that your users experience no downtime at all: for example, you can tell PodSpace to deploy the new Docker image one by one, by stopping an old container, starting up a new one, and then stopping the second old one, etc.

And in the meantime PodSpace is monitoring the health and readiness of your new containers: no traffic gets directed to the newly started containers until all health checks confirm that they are working properly. If something goes wrong, the deployment process stops and the old version is rolled back before your users notice anything!