Hi!
Just a few lines with the most relevant conclusions from the Dockercon Europe 2017 in Copenhagen (Denmark).
The convention was strongly focused on selling the Docker Enterprise Edition. Docker is now a platform and not a tool anymore. It is extended to the Mainframe world (Docker can now containerize legacy applications running on mainframes) and Legacy applications migration on other platforms, naturally.
Docker EE is based on the pure idea of container orchestration for enterprises. The Continuous Delivery stuff is not behind the curtains, it is the core component of container orchestration.
The idea is pretty good and sets the CI/CD methodologies as standards for SDLC, but this is always the same pattern. They build a platform/tool-set to be sold to software departments and small software companies. They will sell the tool to their clients or business departments (the so-called "last mile companies"). Well, the old known pattern producers/clients.
Kubernetes will be fully integrated into the Docker platform. DC/OS 1.10 is integrating Kubernetes as well. Kubernetes seems to be the most serious candidate as the standard container orchestration tool.
Advances in stability, security and scalability are huge. Many demos in this direction. Security has been largely enhanced. Trusted registries are almost mandatory (signed images) delivering containers that have been previously signed, counting with valid certs. Cloud services have to allow this to go live.
Legacy migration to containers is the focus. Why? To sell the idea of Continuous Delivery as the real optimizer of costs. You save money from painful legacy deployment procedures that you can invest in innovation.
Server-less architectures are relying more and more on transient containers (AWS Lambdas works in this way). Monitor, start up time, statistics, etc. have been the focus of some conferences, setting containers as the standard architectural component in Server-less architectures (e.g. the Apache Whisk project).
Developments with LinuxKit have allowed the creation of ultra-light images with an incredible level of security (fully based on permissions, cert based processes, no running on root, etc). Moreover, they are incredibly fast with up times of 100 milliseconds. Docker integration in Windows is actually an admirable piece of work, running a LinuxKit based VM in the background and already integrated in Powershell. So, forget the old Docker for Windows.
Open Source initiatives are very well communicated and integrated to docker, working together (Moby project, containerd) as proposed industry standards.
Conclusions
Cheers!
Just a few lines with the most relevant conclusions from the Dockercon Europe 2017 in Copenhagen (Denmark).
The convention was strongly focused on selling the Docker Enterprise Edition. Docker is now a platform and not a tool anymore. It is extended to the Mainframe world (Docker can now containerize legacy applications running on mainframes) and Legacy applications migration on other platforms, naturally.
Docker EE is based on the pure idea of container orchestration for enterprises. The Continuous Delivery stuff is not behind the curtains, it is the core component of container orchestration.
The idea is pretty good and sets the CI/CD methodologies as standards for SDLC, but this is always the same pattern. They build a platform/tool-set to be sold to software departments and small software companies. They will sell the tool to their clients or business departments (the so-called "last mile companies"). Well, the old known pattern producers/clients.
Kubernetes will be fully integrated into the Docker platform. DC/OS 1.10 is integrating Kubernetes as well. Kubernetes seems to be the most serious candidate as the standard container orchestration tool.
Advances in stability, security and scalability are huge. Many demos in this direction. Security has been largely enhanced. Trusted registries are almost mandatory (signed images) delivering containers that have been previously signed, counting with valid certs. Cloud services have to allow this to go live.
Legacy migration to containers is the focus. Why? To sell the idea of Continuous Delivery as the real optimizer of costs. You save money from painful legacy deployment procedures that you can invest in innovation.
Server-less architectures are relying more and more on transient containers (AWS Lambdas works in this way). Monitor, start up time, statistics, etc. have been the focus of some conferences, setting containers as the standard architectural component in Server-less architectures (e.g. the Apache Whisk project).
Developments with LinuxKit have allowed the creation of ultra-light images with an incredible level of security (fully based on permissions, cert based processes, no running on root, etc). Moreover, they are incredibly fast with up times of 100 milliseconds. Docker integration in Windows is actually an admirable piece of work, running a LinuxKit based VM in the background and already integrated in Powershell. So, forget the old Docker for Windows.
Open Source initiatives are very well communicated and integrated to docker, working together (Moby project, containerd) as proposed industry standards.
Conclusions
- Docker EE for clients and "last-mile companies". Honestly, I'd prefer a better solution than buy the suite, provide the Cloud Provider and pay the bill. Good for huge migrations of tons of legacy apps? Probably, it depends on every case.
- The most interesting thing here is Open Source projects. The Moby project, ContainerD are really interesting initiatives to look around. We need to be expectant about the role of Docker in these initiatives and what is exactly its involvement and dominant role.
Cheers!
Comments
Post a Comment