Docker Best Practises: at what size application should you refrain from containerising?
In my ongoing quest to Containerize All The Things, I’ve played with and installed some pretty hefty images. Not as crazy as what Jessie Frazelle does, but still.
I have one that is 1.8 GB in size at which it hit me whether I may be heading into anti-pattern territory. I mean, where does it end? Dockerize the entire suite of Oracle products?!
Is there an advisory, guideline, best practise guide, anything that states at what point you should just stick with good old servers, VMs or similar?
I ran across a blog that said that SQL database servers is pretty much one where it’s not worth it: by nature they are not immutable and you’re looking at lots of mapped volumes anyway. Naturally I can’t find that blog anymore. 🙁
At one point I ran nearly the entire suite of Atlassian products (Bitbucket, Jira, Confluence, Bamboo) in containers, with massive
docker run command lines to get all the mappings and environments settings in. I stopped doing it when I just couldn’t see the advantage.
Is there a good rule of thumb? A simple formula that encompasses, say, immutability and size to arrive at a “OK to containerise/DON’T containerise” table?