After a sleepless night, I came up with some considerations, and I hope I can start a discussion based on these ideas.
Currently, about three fourths of all desktop computer are using Windows as their operating system, despite serious security problems. Still about one sixth of them are still using Windows 7, although support for this version has been canceled last year. The majority of the Windows systems - about 80 % - are using Windows 10, which poses severe security risks: Each month, Microsoft has to issue a series of patches in oder to mitigate security holes, and all too often these patches create new problems, up to boot failures and inaccessible devices. On the other hand, the many security holes that surface permanently allow severe attacks like infections of Trojans like Emotet, which may ruin the complete installation.
Furthermore, the use of Windows 10 is illegal in the EU, if personal data are processed and the telemetry transmission is not blocked. The GDPR requires that any transmission of personal data to a location outside of the EU is inhibited without explicit consent of the concerned persons. With Windows 10 up to the Professional version, this cannot be guaranteed if the system is connected to the internet, because the telemetry cannot be completely switched off. Especially small installations, like for instance doctor’s offices, have no choice of buying the LTSC (enterprise) Windows version which allows to disable telemetry completely, and mostly they have not the money and the knowledge to protect their systems adequately.
But still these systems are used for a lot of sensitive data processing, mostly because they are dependent on application software which is available only for Windows 10. Replacing these systems with open source without telemetry is not possible, because their applications are simply not available for Linux. The user therefore just has to select between illegal data processing or no data processing at all. One sad consequence of this situation is that Linux, after all these years, can be neglected as a desktop system, as its market share is still below 2 %.
In this situation, a tool to use Windows in a secure way is desperately needed. Qubes provides such a secure environment for using Windows, if the Windows VMs are installed as pairs of Template VMs and AppVMs. The Template VMs can connect from time to time to the Microsoft servers in order to validate the software licenses for Windows and Office 365, but they do not contain any personal data, and no applications need to be started on them, thus reducing the risk of their compromise considerably. On the other hand, personal and other sensitive data can be kept in their corresponding AppVMs, which should not need any internet access, as applications like mail and web browsing can be performed from Linux AppVMs. Additionally, external documents can be sanitized in dispvms before being passed to a Windows VM for further processing. And if, anyhow, a Windows AppVM should get compromised, any malware that got stored in the system part can simply be removed by restarting the AppVM.
For these reasons, Windows support in Qubes is of high importance. Regarding the relative distribution of Windows and Linux desktop systems, a good Windows support within Qubes may help to promote the number of Qubes installations considerably. Currently, Qubes supports aabout a dozen Linux flavors, which in itself is a good thing, but still a niche market. Providing better support for the mass market, i.e. Windows, would make this fantastic system still much more useful - and the Windows users need all help they can get.
So, what should be done? In order not to overload the core development team of Qubes, who do an excellent job, especially rearding the very limited resources available to them, Windows support currently looks as something deserving community activities.
-
In the short term, it would be helpful, if, based on the very good preliminary version of Qubes Windows Tools (in December it was 4.1-65) an rpm file for installing and testing should be provided and made downloadable from a community repo. This would help all those users who are not able to build QWT themselves, and I suppose these are the majority. (Here I am looking at @elliotkillick, (@Eliot1984 ?) @jevank (@tabit-pro ?), @fepitre, and perhaps some others.)
-
In the mid term - say to the release of Qubes R4.1, a version of QWT should be provided where the current problems (limited graphics support, especially only some screen resolutions; missing start button for Windows 7; perhaps even seamless mode for Windows 10; still some glitches during installation; scratchy audio) are solved. Support for Windows 7 should not be dropped, as this system still can be used securely as a Qubes VM, and for several environments it is still vastly superior to the buggy, un-ergonomic Windows 10.
-
In the long run, Windows support should be as stable as the support for other systems. Perhaps a solution like for Whonix could be found, where the workload is mainly carried by developers outside the main Qubes team. If this might lead to a larger user basis for Qubes, maybe it might also help to acquire some fund for the development. (I am perhaps daydreaming…)
I hope these considerations may start a discussion which could lead to some tangible results, and I am looking forward to hearing from you.
