Our company has a dedicated Linux server that wants to host all services on it. We have several wordpress, laravel, asp and node websites. We want to dockerize all of these. But we want all services to use the same mysql. Should we also run mysql in Docker? or not.
How will it be to up and down Docker Compose of one of the projects? Do they affect each other? I am a little confused.
CodePudding user response:
Well, it all depends on the size of your application/services. On a virtual machine, I would not suggest Dockerizing everything and running a docker-compose
to up services. Take for example a database like MySQL
, in docker container there are some constraints like the maximum size of the volume/container and networking, which by using the docker-compose
you need to take care of with additional parameters, daemon changes. Which can be all configured but to know what exactly needs to be configured in what way is a painful process. There can also be problems with the replication of database, you should not have one database in production. What if the data gets lost? Shouldn't you have a second replica?
Now, for the reverse proxy, it depends. Depends on the size of the production as well. What happends if the container is restarted, upgraded? Will the proxy be down and all your services be unavailable? YES! It may be only for a few minutes, but this is production we are talking about.
On the other hand, it all depends on the size of the project, the size of the traffic, and the budget. Take for example a deployment on kubernetes (you did not specify the deployment target, only docker compose so i will default to kubernetes), where everything is in the form of containers. For each node, you have a ingress-controller
(one of the most popular is nginx). If this is production you are talking about, then you can write ingress rules to route the traffic. Ingress-controller
is deployed as a DaemonSet, so each node has its own ingress-controller
and if one node is down, you would also have another one. The same goes for the database.
What I am trying to say, is that running a simple docker-compose on a machine in production is very risky. Use an environment that can scale up either horizontally or vertically (docker swarm, kube). I hope, I clarified the idea behind the production deployment well.