Orka on Apple Silicon Beta

Orka 2.0 has arrived and with it Orka support for Apple Silicon. This new functionality allows you to deploy Orka VMs on Apple Silicon hosts and gain all advantages that come with this platform.

Architecture

The Orka on Apple Silicon architecture differs a lot from Orka on Intel. The latter runs VMs in a container on a Mac host running Fedora CoreOS.

The former takes advantage of the native macOS virtualization framework and runs the VMs directly on a Mac host running macOS. This provides greater performance compared to the VMs running on Intel hosts.

Feature Parity With Intel VMs

Because the Orka on Apple Silicon architecture is completely different from Orka on Intel, there cannot be a 1:1 feature parity between both. 

Some features like IO Boost and GPU pass-through come out of the box on Apple Silicon hosts. 

Other functionalities such as Image Resize are currently not supported by Apple Silicon, and different functionality needs to be implemented to solve the problem that Image Resize is solving on Intel-based hosts.

Functionality

Orka on Apple Silicon Beta comes with all the functionality needed to run a CI/CD workflow. You can:

• Deploy a VM
• Connect to that VM via SSH or ScreenShare
• Configure the VM
• Save an image
• Create a VM config (template) from that image
• Use the VM config in a CI/CD tool - Jenkins, GitLab, GitHub, BuildKite

Note, the Beta functionality is not limited to these features only. You can check all available features in the Orka documentation.

Performance

When it comes to performance, Orka on Apple Silicon significantly outperforms Intel-based hosts in synthetic and build benchmarking runs. A separate blog post that covers this in deeper detail will be published soon.

Performance increases are not limited to CPU, Memory, Disk IO or build times only. Rather, there are significant performance improvements in both VM deployment time and VM boot time as well.

Deployment Time

Due to the new architecture, VM images need to be on the host for a VM to run. In other words, the image needs to be copied to the host during deployment. 

This means that the initial deployment of a VM using a specific image is slower. It can take up to 5 minutes. However, Orka caches the base image on the host, so all other deployments that use the same image to that host are almost instantaneous (about 4 seconds). This is twice as fast as Intel VM deployments, which take about 8 seconds.

VM Boot Time

As mentioned above, Orka on Apple Silicon takes advantage of the native macOS virtualization framework. Because of that, VM boot time decreases significantly. VMs on Intel nodes take about 30 seconds to boot and for SSH to be available. Together with the 8 second deployment time, this means that it takes roughly 38 seconds for an Intel VM to be usable.

Apple Silicon VMs, on the other hand, boot for 3-4 seconds. Including the deployment time it takes 8 seconds for an Apple Silicon VM to be usable. This is already a 30 seconds improvement to build times from an automated CI/CD system that spins up VMs on demand.

Limitations

The most notable limitations are that:

• Currently, only two VMs can be deployed per Apple Silicon node
• Apple Silicon VMs can only run macOS Monterey

In addition, features like Shared Attached Disk and Custom Port Mappings are still being developed and will be released in a future version of Orka.

Sign Up Today

If you are interested in Orka on Apple Silicon, contact us.