In this lightboard talk, Ben Corrie walks through the benefits of containers. For more information, please visit the VMware’s Cloud-Native Apps website: Cloud Native Apps.
In this lightboard talk, Patrick Daigle provides an overview of the networking concepts for vSphere Integrated Containers
Last year we introduced Project Bonneville. The idea behind it, at the high level, is that there is a strong parallel between the constructs Docker uses inside a Linux Docker host and the constructs ESXi uses as a hypervisor. In the final analysis what project Bonneville allowed you to do is to run a docker image as a VM on top of a hypervisor (as opposed to just as a container on top of a Linux host). This has the intrinsic advantage that you can operationalize Docker with the constructs you know and love.
One of the biggest problems IT is facing right now is that their internal customers are asking for “big Linux VMs” only to find out weeks later that they have deployed containerized applications inside those instances. IT has no idea of how to manage, monitor and secure those applications. The Bonneville approach fixes this problem by instantiating those applications as separate virtual machines. Maybe not cool, but very useful.
Fast forward 18 months, we are releasing (and fully supporting**) these technologies as part of vSphere.
Enterprise Plus customers have now the option of leveraging a feature of vSphere called vSphere Integrated Containers (VIC for short).
vSphere Integrated Containers is comprised of three different technologies. What makes them unique is that they are all open source. This means that you can just “consume” what we are building or you can also contribute (if you wish so) features that you may deem as necessary for your particular use case. These three technologies are discussed below.
Note that there is a video at the end of this post that will show these technologies in action. In the meanwhile, this is a 33.000 high level diagram of how these technologies relate to each other:
This a complete rebase of project Bonneville. When the engineering team was tasked with the need to productize Bonneville they decided to re-write it and include a so called Portlayer. The Portlayer is an interface that exposes vSphere objects and services as containers primitives. On top of Portlayer you can have multiple different personalities. As part of the first announcement we have created a Docker personality (think about VIC Engine today as a Docker “façade” on top of vSphere).
The way you create this “façade” is pretty straightforward: as a vSphere admin you will use a tool called vic-machine (which is part of the VIC Engine binary) to deploy a Virtual Container Host (a vApp) on top of vSphere.
Inside the Virtual Container Host there is a small VM that acts as the Docker Endpoint. The IP of that VM is what the vSphere admin will hand over to the internal customers that need Docker. When the customer run “docker run –H
The VIC Engine Github repo is located here.
While one could see VIC Engine as being the core component of vSphere Integrated Containers, we soon realized that Enterprise customers were asking for more. Hence we decided to create a product that would do more than just mimic the behavior of a compatible Docker Engine.
For this reason, vSphere Integrated Containers also ships Harbor, an Enterprise Docker registry. For vSphere Integrated Containers deployments we have bundled it as a virtual appliance in OVA format. vSphere admins will grab the appliance and import it into the vSphere environment.
vSphere admins can then hand off its FQDN or IP address to their internal customers. They can then use the registry service provided by Harbor as a secure Docker registry instantiated inside the data center. Not only they will continue to push and pull to and from Docker Hub, but they now have the possibility to push and pull to and from a local registry.
Harbor is built on top of the open source Docker registry foundation and we added features that most Enterprise customer are asking for: LDAP/AD support, role based access control, a user interface and image replication to name a few.
If you are interested in understanding more about the internals of Harbor this is a good blog post from the engineering team that gets into some of the details.
This is the public Harbor repo on Github. For people that are interested in joining the Harbor community (as opposed to just use it as part of the supported vSphere Integrated Containers product), feel free to interact directly with the engineering team over there and/or submit PRs.
Admiral is an extension of vRealize Automation 7.2 and it adds container support to vRealize Automation. You can find additional information about it here.
However, given Admiral has been developed independently and can be instantiated standalone, VMware decided to add Admiral to the vSphere Integrated Containers product.
Given that with VIC Engine we are leveraging the very robust vSphere features to schedule “ContainerVMs” on top of hypervisor hosts, we are not leveraging all the capabilities that Admiral provides in a scenario where you are using Linux Docker hosts on top of which you instantiate containers. However, we leverage a lot of Admiral features in the context of vSphere Integrated Containers including providing a user interface for Virtual Container Hosts consumption and the capability of composing multi-container applications to be deployed as a single entity.
You can access the public Admiral Github repo here. As a reminder, Admiral is still considered Beta as part of vSphere Integrated Containers.
See vSphere Integrated Containers in action
Now that we talked about the technologies that comprise vSphere Integrated Conatiners, it is time to see them in action. This video shows how to use together the three technologies discussed above.
** Admiral has not been GAed yet so support for Admiral, as part of vSphere Integrated Container, is limited to the level of support we provide for Beta software.