Typical Speedtest Result Expectations for CBK40 Kit

Furry 

Beyond the fact the article was relating more about MB/s and Mb/s, it does little to discuss encoding and transport losses, but even with transport and encoding losses good hardware still acheives a significant portion of the link speed. On unidirectional connections (even in half duplex systems), an 866Mbps connection should yield 866Mbps throughput minus any protocol, encoding and packet loss. With WiFi, this will also be dependnent on many factors, SNR, walls, adjacent channel interference, etc. Rule of thumb is tossing 50% for overhead. Great.  But the reason it is called AC2200 is because that is the combined established connection rate, and, when you consider the AC2200 spec actually calls for sharing a 2.4GHz and dual 5GHz channels to acheive higher throughput, dual 5GHz 866Mb/s are supposed to provide 1733Mb/s, add the 433Mb/s at 2.4GHz, the CBR40 issue is even worse. I'm simplifying, but you get the point, I hope. One of the hidden gotchas with the Orbi CBK40 is it dedicates one of the 866Mb/s channels for backhaul. Which makes it function as only AC1300, not the specified AC2200. It's like selling a 20 core processor but not telling customers 8 of the cores are dedicated for internal use and not available for customer applications.

As an aside, if what you are saying about 5GHz maxing out at 500Mb were true, there'd be no point having an AX4200 or AX6000 specs for 5GHz access points.

As for the wired end: direct connection to the cable modem via wired ethernet, there is zero excuse for the internal CBR40 (with internal modem, router, and switch) to provide less than the established cable speed. So some factor inside the CBR40 is throttling traffic: while the CBR40 can train and establish a 1Gb cable connection, it seems it can't support the traffic throughput (at least based on your experience and mine). As a further example: when run with a 200Mb/s trained connection, the device provides 200Mb/s. Yet, trained at 1000Mb/s, it provides a typical 500Mb/s (again, talking wired). 

I guess I am done with this topic, too. Another example of Netgear marketing outperforming the actual product being sold. 

ps: I can keep a 10Gb pipe full for a while. Intraserver data traffic moving terabytes around. RAIDs and such. Looking at 40GbE now...

---

FURRYe38 wrote:

If your seeing 500Mbps on wireless then this would be with in the spec. Again, wireless will not ever see anything near 900mpbs on 5Ghz AC. The connection rate of 866Mbps is only a connection rate between the wirless devices, if it supports that rate and the RBR/RBS. This differs from actual speed thru put. Don't confuse them. Connection rate differs from actual bandwidth:

https://kb.netgear.com/19668/Link-Rate-and-Transfer-Speed

 

Also antenna support on wirless devices will impact and limit how wireless speed performances work. 2x2 will cap out at around 400-500Mpbs. 4x4 and higher allows for higher speeds. My iphone 12 max is 2x2 supporting, though AX mode supporting, it's stil only max's out around 700mpbs on my Orbi AX system. 

 

Make no mistake, the CBR40 when I had it was working well at the time. This was back in 2017. Speeds were wired and on a 1Gb network, worked just fine. 

 

10Gb, well thats fine but thats only on the LAN side. If you have that then cool. but your bottle neck will be any router or modem with 1000Mpbs WAN and LAN ports which the CBR was only designed for. NG has a few 2.5Gb WAN port supporting devices. 10Gb hasn't been seen much yet on a consumer level and isn't quite fully appeared yet. Seems like some mfrs are slowly graduating getting pasted 1Gb and just breaking into 2.5Gb rates. Not alot of consumer class devices can handle 10Gb bandwdith. There is HW considerations to take in when dealing with 10Gb. Thats a spendy speed to put in when you can't fully take advantage of it. Hope you can. 

 

---