Thanks for sharing.
Iâm not disagreeing, not being argumentative, but just informing technically.
There is a form of local-only storage which requires a server for network reasons. The domestic LAN device is opening a port to a WAN server and you access that same WAN server to then connect back to the LAN device. This is to bypass your routerâs firewall and overcome NAT.
Eufyâs alternative approach would be to make the device uPNP and open up a firewall port and then cache the WAN IP on your device. That would make you more vulnerable to anyone who knew your IP and knew you had this device. To secure this arrangement would involve hardware encryption at the device, typically battery operated, at a battery life cost.
Itâs a similar issue with SIP VOIP Proxy.
Not making an argument pro / con , just educating.
Now the true issue is there is nil inherent reason for Eufy to cause an outage of this magnitude. Yes you do need to have outages. Yes you do need to impact users, but you do not need to impact for a long period (above about a minute) and you do not need to impact more than a small fraction of users.
A software upgrade can be done via DNS âround robinâ where a small fraction of users are directed to a new software version, you test stability, and after a while you upgrade. This is often called âcanaryâ CI/CD. You can do it at any level, web, app, DB.
There is no need for an outage either. You can route all traffic to alternative region, then upgrade your primary region, test (see above) then route back. Youâd impact performance during this and some users would experience a short ( order of 1 minute) unreliability during the switch.
For this scale of outage, takes either a particularly inexperienced person, or a focus on cost avoidance, as each of the above outage-avoidance comes at a cost. A small fraction of users, for a brief time, having an issue is a sign of testing done with minimum impact.