It seemed like today was all about the cloud and cloud native apps. Among the many announcements, VMware announced two key new capabilities: VMware integrated containers and the
Python Photon Platform.
Containers running on VMware
- VMware vSphere Integrated Containers is an implementation of containers that runs natively under vSphere. The advantage of this solution is that now when developers fire up a multi-container app, each container now exists as a separate VM under vSphere and can be managed, monitored and secured just like any other VM in the environment. Previously a multi-container app would be one VM per container engine containing potentially many containers running under the single VM. But with vSphere Integrated Containers, the container engine and the light weight Linux kernel (
PythonPhoton OS) are now integrated into the ESX hypervisor so each container runs as a native VM. Integrated containers is an follow on to a combination of Project Bonneville, Project PythonPhoton (OS) and Instant clones. Recall with Instant Clones one can spin up a clone of a VM in less than a second and its memory footprint is 0MB. PythonPhoton Platform takes container execution to a whole new level, with a new deployment of a hypervisor tailor made to run containers (not VMs). With the PythonPhoton Platform one natively runs container frameworks underneath the platform. PythonPhoton Platform consists of PythonPhoton Machine which is PythonPhoton OS (lightweight Linux Kernel distro) & the new Microvisor (new light weight hypervisor for container hardware calls) and PythonPhoton Controller which is a distributed control plane and management API. With PythonPhoton Platform one can manage 100K to Millions of containers, running under 1000s of container frameworks.
Python Photon Platform is intended to be open sourced. VMware also announced a bundling of Pivotal Cloud Foundry with the Python Photon Platform so as to better run cloud native apps implemented in Cloud Foundry. But the ultimate intent is to provide support for Google Kubernetes, Apache Mesos and any other container framework that comes out.
So now you can run your Docker container apps or any other container app solution in two different ways. One depends on vSphere standard management platform and runs container apps as a standard VMs. The other takes a completely green field approach and runs container frameworks natively in a ground up new hypervisor solution with a new management solution altogether that scales.
The advantage of
Python Photon is that it scales to extreme, cloud level types of application environments. Python Photon is intended to run cloud-native apps.
vCloud Air extensions
One of the other major things that VMware demoed today was moving a VM from on premises to vCloud Air and back again – a real crowd pleaser. One VMware Exec said that after MIT had convinced them they needed to be able to move apps from on premises to the cloud for dev-test apps. They then turned around and decided they wanted to move dev-test activity back to their onprem environment and instead wanted to move their production to vCloud Air.
They demoed both capabilities using vMotion to move a VM to vCloud Air and using it again to move it back. The nice thing about all this is that all the security and other attributes of the VM can move to the cloud and back again along with the VM. All the while the VM continued to operate, with no disruption to execution. They mention that it could potentially take hours to move the data for the VM.
There were a number of other capabilities announced today including EVO SDDC (EVO: RACK reborn) which includes a new datacenter management solution. Customers can now roll in a rack of servers and have EVO SDDC manage them and deploy software defined data center on them in a matter of hours. Within EVO SDDC you can have application domains which span racks of servers but provide isolation and management multi-tennancy.
NSX 6.2 was also discussed and essentially is key to extending your networking from on premises to vCloud Air. With NSX 6.2 local routing, micro segmentation security and app firewalls can be configured locally and then be “extended” to the vCloud Air environment.
Lots of moving parts here and I probably missed some key components to these solutions and didn’t cover any of them well enough other than to give a feel for what they are.
But one thing is clear, VMware’s long term strategy is to take your native, on premises VMs to vCloud Air and back again as well as if your Dev-Ops group or any other BU wants to use containers to implement cloud apps, VMware has you covered coming and going.