As other answers have confirmed, separating router and wifi AP into discrete devices can provide advantages due to:
- better placement options for the AP
- reduced CPU workload
- more discrete feature management than many integrated devices allow
- more diagnostic and troubleshooting tools available in the equipment
However, it also requires a much more intentional network setup and design compared with an integrated unit in order to avoid common pitfalls.
Anyone wanting to take advantage of these advantages needs to put their ego aside and understand that they are signing up for a higher knowledge and skill commitment vs either an integrated unit, or routermodem + big box home wifi router.
In short, the advantage is you can do all the things your friend hasn’t (reportedly) done to diagnose and resolve issues such as the one your friend is having!
“Tech guys” in my experience more commonly run into the pitfalls inherent to ANY discrete router+AP setup when using Ubiquiti devices vs Mikrotik or Cisco (two other most common brands I see used both internally by large businesses and by enthusiasts) because Ubiquiti documentation and support is nowhere near so accessible, universally applicable, and detailed as the other brands.
With Ubiquiti (or for that matter Google router or Orbi or similar “mesh” kits) the cost and knowledge barrier to entry is lower so you can make a half-working kludge and get away with it for a while. You must remember both the Ubiquiti AP and the Google device he replaced it with may not be a true AP!
”Layer two switch with wifi capability” is what many people think they mean by “AP” but it is not inherently so unless you configure it that way.
Devices of this class tend to have built in router functions too, but with the Ubiquiti EdgeRouter filling that role the “AP” must have its routing functions disabled to act as a layer 2 switch with a layer 3 management interface.
When you separate router and AP, you must also inform the devices that they are to play separate roles.
My day job is to support this setup (on a WISP scale not an in-home scale) and I can tell you that unless it’s a model that doesn’t know how to be a DHCP server and DNS proxy, or unless he has disabled these features in the “UniFi”* the odds are that it is competing with the EdgeRouter for this role and the network may actually be double-NATed or worse experiencing rogue DHCP. The Google router may be similarly confused.
- Footnote: as pointed out in comments, the true UniFi AP doesn’t have a DHCP server built in but other products from the same manufacturer say “UniFi” on the login screen. YMMV both with the DHCP and with actually determining what Ubiquiti really calls that particular device.
Your friend has apparently ignored the other massive advantage to discrete devices: separating variables while troubleshooting
What he (and anyone who is similarly tempted to throw new expensive equipment at a problem and expect it to automatically fix it) should do is attempt to replicate the problem one device at a time. This means starting with eliminating the wifi leg of the connection and testing from a wired device.
The EdgeRouter (and any discrete business router) also allows things a big-box router usually can’t such as looking for Ethernet negotiation speed and checking MAC and ARP tables. This is essential as well. Most likely if the routing vs switching configurations are correct he is having a layer 1 problem such as poor wifi modulation/interference/fade, or CRCs on the cabling (especially if the cable runs outdoors or if he custom terminated it).
The discrete AP unlike most integrated units is generally capable of spectrum analysis so instead of guessing at the best channel, you can scan and see. This is excellent when planning multi-AP buildings like your friend has seen using these.
The discrete devices are also capable of pinging and tracerouting from within themselves. Therefore he could investigate the DNS issue from within the devices while it is occurring.