I would really like to hear from Docker about this as well. I’m worried about where Swarm is headed. I really like it, but I’m not sure what is going to happen to it in the future.
I would be also happy to hear something about the swarm roadmap, since we are developing right now a larger software package on swarm. If we have to move within the next say 5 years to Kubernetes, it would be a huge effort and wuold raise big questionmarks…
@bitsofinfo
There was some time ago an explanation from a docker stuff member, that they have sometimes a look around, but they don’t really participate at the discussions as long as they don’t think, that it is really important to do.
So don’t assume any reply from docker itself. If you want an answer, I would assume, that you have to use Docker EE and open a ticket.
Swarm is alive and well. There’s no roadmap that I know of that’s kept up to date. The docker/swarm links you refer to is Swarm Classic, a solution used before SwarmKit was released in 1.12 (July 2016) that replaced the similarly named Swarm repo. docker/swarm is now deprecated in favor of this repo (aka Swarm Mode, or just Docker Services). Lots of Swarm fans at DockerCon.
So true, there is no public roadmap for SwarmKit, though one could argue you might look at recently updated PR’s but that’s not a guaranteed list of future features.
If you want to learn what’s changed recently, you can look at the Docker CE release notes… looking for anything Swarm Mode and Overlay Networking related.
Also of note, as Docker Captains, we were lucky enough to sit with various Docker Inc. product managers At DockerCon 2018 and they were all committed to the future of SwarmKit/Swarm Mode, saying that customers loved it. At this point I think it’s just a question of engineering priorities across the Docker toolset and bugfixes vs. new features… but I can’t speak for Docker Inc. directly
@stephanwalter 5 years are an era in our field: it’s impossible that a technology can remain the same, or even survive during this period. If you want to pick a technology to use for the next 5 years…well, you probably need to do it yourself, or just pick the best that you can find around and hope that it will not die during this period.
I’m a longtime swarm (swarmkit - docker services) production user. I started using swarm, and I continue to use it happily, because it’s the uncomplicated orchestration software that balance well features, easiness (abstractions) and flexibility. With it you can have a production level cluster in minutes: have you ever tried setting up a production level kubernetes?
But, obviously, this is only my microservice-based-ecosystem-case, and luckily docker swarm fits well it this environment. In my situation I can replace it with many similar solution without too much hassle because it just provide private network, service discovery and replicas distribution between nodes.
I read your comments in the other topic about swarm future: if you need to build your entire business on swarm probably you need to become an EE customer, or rely on another company services. Because, if you are an open source user you’re always on your own, and this is true with any open source software (nobody works for free). Yes, you can have support from the community but the community doesn’t work for anyone.
For my part, I stopped worrying about “docker swarm future” because:
it’s here and it works
nobody from Docker Inc. said that it will be deprecated
the roadmap isn’t a real problem because I think that Docker Inc wants to keep it simple and add only small/medium improvements during the year
if I need more than this I will just switch to another software (or service)
For me the fact that the experimental API adds the notion of managed plugins deployed to the swarm is a very big thing as it will eventually simplify my environment build scripts to simply do base Docker rather than adding the managed plugin step to the provisioning scripts.
Looking at https://github.com/docker/docker-ce/releases it seems that Swarm items are only bug fixes, but not any new features for a while. So we likely won’t get things like Auto-Scaling in the near future.