Today, Canonical released Anbox Cloud which builds upon the same ideas as I built Anbox on in 2017. I am thankful that Canonical allowed me to take the idea of Anbox to the next level, form an engineering team around it and bring it as a fully supported product to any cloud (private or public, thanks to the magic of Juju) for a wide variety of use cases. It covers more advanced use cases like cloud gaming or application streaming for enterprises down to providing an on-premise Android cluster for development and testing of Android applications. You can read more about Anbox Cloud here, here and here.
Despite the nature of Anbox, Anbox Cloud is not open source and a commercial product.
It is based on the same underlying ideas of Anbox but is a completely separate code
base. However that doesn't mean my team at Canonical isn't contributing back to the
original Anbox. We have worked for example very hard (especially
Christian Brauner)
to bring binderfs
to the upstream Linux kernel. It is now enabled in all new Ubuntu
kernels (>= 19.04) by default and can be loaded at runtime if needed
($ sudo modprobe binder_linux
). Anbox itself supports binderfs since this PR.
This will allow Anbox to work out of the box on any Linux system without the need
to load custom kernel modules which are not part of your distribution kernel. I expect
binderfs to come to more distributions by default in the near future. From
recent changes to the Android
Open Source Project (AOSP) it looks like Google will make use
of binderfs in upstream Android starting with Android 11 too :-)
Anbox itself will remain as is, licensed under the terms of the GPL-3 and a community project centered around the Linux desktop, aiming to provide an Android application layer (like WINE) on any Linux system.
Thanks to everyone who contributed to Anbox so far in one way or the other and those who will in the future!
We have an exciting future ahead of us!