Rocket vs. Docker and the Myth of the “Simple, Lightweight Enterprise Platform”
1–10 of 21 posts
Re: Rocket vs. Docker and the Myth of the “Simple, Lightweight Enterprise Platform”
#2I cant read this whateveritis because the fonts are too gray/white as the background.
Skimming through it, yes whatever you define as "enterprise" can fit your definition and whatever claim you are doing in the article or blog, can be fulfilled.
Re: Rocket vs. Docker and the Myth of the “Simple, Lightweight Enterprise Platform”
#3I cant read this whateveritis because the fonts are too gray/white as the background. Skimming through it, yes whatever you define as "enterprise" can fit your definition and whatever claim you are doing in the article or blog, can be fulfilled.
Re: Rocket vs. Docker and the Myth of the “Simple, Lightweight Enterprise Platform”
#4??? Does the author understand ANY motivations for rocket? They're asking for exactly what rocket has set out not to be.
Rocket doesn't contain an orchestration platform, become rocket knows it's a tool to be composed into an orchestration system.
It will work excellently with Kubernetes.
Re: Rocket vs. Docker and the Myth of the “Simple, Lightweight Enterprise Platform”
#5He's setting up a nice strawman. I didn't see anything in the Rocket proposal that implied they thought the things Docker wants to add aren't needed, just that they object to the direction and want to be able to pick and choose from components that fit their purpose better, and want guarantees of interop through open formats etc.
Who is propagating this myth of the "Simple, Lightweight Enterprise Platform" to begin with? Some searches did not net me any references to that other than this blog post.
Re: Rocket vs. Docker and the Myth of the “Simple, Lightweight Enterprise Platform”
#6You've got your conclusions backwards. Docker is the one that's trying to do things the simple way. Rocket embraces the Unix process model and philosophy. They don't have to solve this particular security problem because it's been solved over almost 40 years of Unix evolution.
The Docker simplicity is why they've had so much success. That's how it started: as a tool to make using LXC simpler.
Re: Rocket vs. Docker and the Myth of the “Simple, Lightweight Enterprise Platform”
#7It happens every time.
As soon as you hit an environment where there are > 20 people using the system you need a plethora of tools and systems involved which remove the terms, "simple" and "lightweight," from the specification.
Access control, authentication, auditing, operations... you can't eschew complexity by ignoring it.
I've learned this time and again. If you start out with "simple" and "lightweight" you're going to hit that "enterprise wall" and really pay the price of not considering those requirements up-front. The price of either re-writes or layers of out-of-band processes and tools for orchestrating light-weight systems that were not designed to be composed in such a way. It's quite painful and terrifyingly complicated.
Update: In the case of Docker/Rocket... I think these are just early iterations. I wouldn't be surprised if even Rocket becomes bloated as they realize they cannot design for simplicity when complexity is a requirement.
Re: Rocket vs. Docker and the Myth of the “Simple, Lightweight Enterprise Platform”
#8He's setting up a nice strawman. I didn't see anything in the Rocket proposal that implied they thought the things Docker wants to add aren't needed , just that they object to the direction and want to be able to pick and choose from components that fit their purpose better, and want guarantees of interop through open formats etc. Who is propagating this myth of the "Simple, Lightweight Enterprise Platform" to begin…
> I didn't see anything in the Rocket proposal that implied they thought the things Docker wants to add aren't needed
I was referring mainly to this statement: "We should stop talking about Docker containers, and start talking about the Docker Platform. It is not becoming the simple composable building block we had envisioned."
I agree with you that that doesn't necessarily mean that the Rocket team disputes the need for enterprise features, but (as you say) may think other technologies are better suited. The announcement itself doesn't go into much detail as to which technologies those would be, though.
> Who is propagating this myth of the "Simple, Lightweight Enterprise Platform" to begin with?
It certainly hasn't been associated with this specific phrase, but the idea that there is a New Cool Tech around the corner that will be simpler, easier and less heavyweight than the Last Cool Tech seems to be a recurring theme in our industry, in my experience.
That's not to say that new technologies don't introduce dramatic improvements in certain areas. Personally, I'm a big fan of microservices and, specifically, containers as a possible implementation mechanism.
Equally, I don't think it's surprising that technologies such as Docker that start out as 'simple, composable building blocks' become part of more complex, "heavyweight" solutions as they evolve.
Re: Rocket vs. Docker and the Myth of the “Simple, Lightweight Enterprise Platform”
#9It happens every time. As soon as you hit an environment where there are > 20 people using the system you need a plethora of tools and systems involved which remove the terms, "simple" and "lightweight," from the specification. Access control, authentication, auditing, operations... you can't eschew complexity by ignoring it. I've learned this time and again. If you start out with "simple" and "lightweight" you're go…
> I wouldn't be surprised if even Rocket becomes bloated as they realize they cannot design for simplicity when complexity is a requirement.
That was pretty much what I was trying to get at with the question of how a "Rocket ecosystem" would address these requirements. I think vidarh makes a valid point that these could be met by tools outside the container system that are better suited...but I am curious as to what this combination of tools would be.
Re: Rocket vs. Docker and the Myth of the “Simple, Lightweight Enterprise Platform”
#10You've got your conclusions backwards. Docker is the one that's trying to do things the simple way. Rocket embraces the Unix process model and philosophy. They don't have to solve this particular security problem because it's been solved over almost 40 years of Unix evolution. The Docker simplicity is why they've had so much success. That's how it started: as a tool to make using LXC simpler.
> The Docker simplicity is why they've had so much success. That's how it started: as a tool to make using LXC simpler.
Do you feel Machine, Swarm and Compose add to or detract from the value proposition of Docker, in that case?
Just to ensure I'm understanding your point about Rocket benefiting from 40 years of Unix evolution correctly: do you think that Rocket will be able to remain simple as a result of this? And that Docker perhaps will not, even though up until this point it's been "trying to do things the simple way?"