Does Azure Service Fabric do the same thing as Doc

2019-02-01 13:22发布

My thinking is that people use Docker to be sure that local environment is the same as production and that I they can stop thinking about where are their apps running physically and balancing mechanisms should just allocate apps in best places for that moment.

I'm 100% web based and I'm going to move to cloud together with our databases, and what cannot be moved will be seamlessly bridged so the corporate stuff and the cloud will become one subnetwork.

And so I'm wondering, maybe Service Fabric already does the same thing that Docker does plus it gives as address translation service (fabric:// that acts a bit like DNS for the processes in fabric space) plus (important for some) encourages on demand worker allocation - huge scalability perk.

  1. Can Service Fabric successfully replace Docker?
  2. Is it gaining audience and acceptance? Because otherwise even the greatest invention can fail.

3条回答
等我变得足够好
2楼-- · 2019-02-01 13:39

The key similarities between the Service Fabric and Docker containerization:

  1. Both Dockers and SF, are capable of creating an immutable image out of your micro-service implementation, on both the platforms - Linux and Windows.
  2. Both Dockers and SF, are capable of orchestrating your containerized application within a cluster of VMs. These VMs can be anywhere - public cloud, private cloud or your own data center. Please note that both of them are cloud platform agnostic, that means, they don't have strong affinity on any of the cloud service. So as long as you are not using any cloud specific feature within your micro-service, this should be fine.
  3. Both Dockers and SF, are capable exhibiting essential capabilities of an orchestrating platform: Service Discovery, Service level load balancing, Network level isolation among services, fail-over handling and replication control etc.

The key differences between the Service Fabric and Docker containerization:

  1. Docker container is essentially a deployment / packaging construct. That said, docker doesn't dictate on what you are packaging within a container as part of your service implementation. Neither it provides any programming construct to implement your kind of service. Whereas, Service Fabric provides programming constructs in the form of base types / interfaces from which your service implementation can start with a certain kind of service declared - stateful service, stateless service, virtual actor.
  2. In the Docker world, everything is a container, i.e. your minimum deployment / orchestration unit is a container. Hence, it doesn't recognize or support an individual process. Whereas, in SF, we have a provision wherein, your micro-service derived from stateless / stateful service can be orchestrated and governed as a process. However, SF also supports container orchestration the way Docker does. Also, the latest version of SF allows packaging your stateful / stateless service within a container.

With above facts in mind, please note that SF doesn't have any strong affinity on any cloud provider. It can run equally on any public cloud - Azure, AWS or GCP, as long as you are able to create the VMs with desired platform.

查看更多
三岁会撩人
3楼-- · 2019-02-01 13:52

It is not comparable at all. With service fabric, you get health monitoring, code integration with the fabric, logging, monitoring, load-balancing, and other intelligent features. Your application can even execute shutdown code. Service Fabric is not just for Microsoft technologies and even docker can reside inside SF so is rkt or Unix OS. Security and networking features(in-line with web apps) is another plus. Reliable collections is simply brilliant. And a roadmap to better application building and performance is guaranteed for the companies adopting it (history says so).

This question is highly favoring 'greatest invention' Docker. This comparison can do good for Docker marketing but no one will replace SF for Docker. Docker is just a tiny OS copy (nothing to do with services, applications or intelligence). Docker even has nothing to do with application development, it wasn't the intention. Just that people have started to find the need for isolation and sharing. And that is what Docker is all about.

查看更多
Rolldiameter
4楼-- · 2019-02-01 14:00

It's confusing since Docker (the company) is trying to stake claims in everything cloud.

  • Docker Engine (what most people call "Docker") is a containerization technology. It can give you
    • Process isolation
    • Network isolation
    • Consistent application environment
  • Docker Hub is an image registry. It stores Docker images so you can download them as part of your deployment.
  • Docker Cloud is an orchestration system for Docker. It can give you
    • Scale your applications up and down
    • Connect your applications to each other
    • CI testing, integrated with Docker Hub (this isn't part of orchestration, just another thing it does)

Service Fabric is an orchestration system. It can orchestrate Docker containers, but it can also integrate more tightly with your services if you build specifically for Fabric. (Docker is completely agnostic about what runs inside a container.)

So Service Fabric is mostly comparable to Docker Cloud, though it's not an exact match. There are some other Docker-based orchestration solutions (Kubernetes is probably the biggest) and there are other cloud-based micro-service solutions (Heroku is probably the best-known).

The primary disadvantage of Service Fabric is that it's a Microsoft technology and so you're going to be tied to Azure to a greater degree than if you were running Docker. The other is that Docker has a broader range of choices for building your stack: all three Docker-things I listed above have at least one open-source alternative (this is also a big disadvantage of Docker, since nobody's laying out a single Best Practices For You document).

If you love Microsoft and if cobbling systems together is not something that's important to you, then Service Fabric should be a fine alternative to the Docker ecosystem. (And you can still run Docker containers under it.)

查看更多
登录 后发表回答