No, I’m not saying that there’s no difference between Linux, which allows for a speed test of 1Gbit/s, and Mirage, which has less. I tried to explain that I suppose this is due to a feature that was not implemented in Mirage (I only suppose because if I turn off the offload on Linux I have similar bandwidth with Mirage). But as a daily driver I can’t see a difference, e.g, for video meetings, mails, and web access, I don’t notice network lags (for video meetings, the bottleneck seems to be my firefox AppVM browser), nor 100%CPU usage for mirage.
Yes, by others, I mean other users of the Mirage firewall. Personally, I don’t want to target anyone nor force anyone to use Mirage, but I always try to listen to requests for features or bug reports. My limited spare time on the project allows me to fix some issues, but unfortunately not all of them.
Recently, some users on this forum have indicated that they would like to test Mirage, but find the installation procedure too complex or daunting. The solution seems to be to produce an RPM with Qubes-builder, but I don’t have the time to improve my skills with this tool (however, if anyone else would like to do so, I can assist, please anyone, come here
).
Regarding speed tests, it’s on my to-do list, but not a high priority, because I feel the improvement is mainly noticeable for this measurement tool, or for downloading large files (excluding torrent-type protocols).
Other improvement on my to-do list are IPv6 support, and multicore support.
And finally, I try to improve bandwidth for qubes-mirage-ovpn (a OpenVPN tunnel unikernel), and qubes-mirage-dnshole (a PiHole inspired unikernel).