I used docker swarm while exploring about orchestration and got to know about docker swarm it is very easy to setup and ready to deploy with minimal configuration. Review collected by and hosted on G2.com.

While I was using docker swarm I observed its documentation not clarifying all things in attractive way. Also realized they ca n improve in load balancing services like ingress. Review collected by and hosted on G2.com.

Zero-downtime deployments are straightforward to configure in the service. The swarm will not send traffic to the updated container until it is healthy and will not shut down the old one until the pending request has finished. Also, if a service doesn't start correctly (health check failure), Docker is smart enough to keep the previous version running.

The configuration file (docker-compose.yml) is idempotent, so only the changes to the service are applied when deployed. I tried to learn k8s, but it had many moving parts that make debugging very hard when something goes wrong, compared to Docker Swarm, where everything uses the same daemon in every node, so the log is centralized.

I love that the service description in the config file can be straightforward because many things are implicit by design (like network/volume names) or easier to see. For example, opening a port for a container in every node can be done in a single line. Review collected by and hosted on G2.com.

Over the years, I found some limitations when deploying services on the Swarm. Recently they are solved, but most have opened issues in GitHub for many years. Some examples:

* Opening a shell to a container requires locating the node where the service is running, then logging into that node.

* Cannot access special devices (like a GPU) from my containers running in Swarm mode without hacks.

* I cannot update secret/config in services without changing their referenced name in the service config. I had to create a tool to overcome this limitation.

There are tools that help some pain points (like Portainer), but one shouldn't need to rely on 3rd party utilities to handle the software.

When testing Kebernetes, I found that most of these issues are solved or don't exist in k8s. Review collected by and hosted on G2.com.

It's easy to setup a cluster with very minimum things. It gives flexibility to setup ,Routing and things in ingress very easy. It makes orchestration easy way to production ready. Review collected by and hosted on G2.com.

I also used kubernetes so observed some differences like docker swarm need to improve document, networking compatibilities and health checking .It was mess sometimes when I were using. Review collected by and hosted on G2.com.

Easy to configure and use and it is fast Review collected by and hosted on G2.com.

There is no auto scaling feature ....... Review collected by and hosted on G2.com.

As we all know, Docker is a containerized engine that we use to create and manage containers. To automate processes, we use Docker Swarm, which can be considered similar to or an older version of Kubernetes. After Kubernetes entered the market, the value of Docker Swarm decreased, but in some scenarios, Docker Swarm is still preferred. When we talk about orchestration, it means coordinating like a musician or singer orchestrates their group, deciding when the drummer should play, what the guitarist will play, and so on.

In a containerized environment, it is used to orchestrate images. If the images stop or crash, what is the policy to revive them, and how to create duplication or replication of them? This is how it is used and helps automate the handling of images for a DevOps person. Review collected by and hosted on G2.com.

We can only orchestrate Docker images, but in Kubernetes, we can take any containerized engine and orchestrate them all together. Review collected by and hosted on G2.com.

Easy to learn, very simple to use. Can be easing integrated with docker. You can add n number of hosts in the swarm and be able to auto scale the instances based on the volume Review collected by and hosted on G2.com.

Does not have all the features like kubernettes however with kubernettes there is a learning curve and overhead of using additional tool so if you don't have large volume I would go with docker swarm Review collected by and hosted on G2.com.

The deployment process is a little fast in Docker Swarm as compared to any other container orchestration tool, and also this is free of cost Review collected by and hosted on G2.com.

We will need to manage everything manually, and also auto-scalable is not possible in this. Apart from this will not manage taint and tolerance Review collected by and hosted on G2.com.

It comes along with docker community editions, no need for additional configuration to enable it.

1. Service discovery - It supports service discovery; if you need more, you can use traefik or istio on it.

2. Scaling - Supports auto-scaling

3. Nodes - adding additional nodes is straightforward

4. Volumes - Volume mounting is stragightforward

5. Logging - Support prometheus, grafana, elk etc

6. Support - You get good support from docker community

7. Open source - Well maintained docker runtime Review collected by and hosted on G2.com.

Changes need redeployments and costly processes compared to Kubernetes orchestator Review collected by and hosted on G2.com.

Easy to install

Docker having it's own tool for managing all containers that's docker swarm Review collected by and hosted on G2.com.

Not having any dashboard for centrally tracking

And it would be great if docker swarm creates container as per need or requirements automatically Review collected by and hosted on G2.com.