Nicki Palmer, Verizon's highly respected Chief Network Officer, spoke of 10 ms latency for 5G Edge. "We have architecture up and running in New York City and we are seeing sub-10-millisecond latency right there as we continue to test," she said in May 2019. Sue Marek reports, "The latency obtained through the company's edge computing test in Houston were 15 ms or half of what's available on Verizon's LTE network." (Verizon 4G often is 30 ms or less, although 4G on many nets should be considered 35-50 ms.)
"Thierry Sender, director of edge computing can guarantee customers latency of between 25 to 50 milliseconds," Sue Marek reports. WTF?
Sender's comment implies has abandoned low latency Edge. It still is adding servers, looking to rent them out to Amazon and (they hope) gaming companies.
However, PCMAG has tested Verizon's network extensively and sometimes did find 10 ms latency. It has one of the best and fastest backhaul networks in the world, adding 1400 miles of fiber every month. It has replaced 200,000 older boxes (mostly routers) with 20,000 much faster units.
I believe 10-15 ms remains Verizon's goal, although 1 ms is a fantasy outside the lab. So I'm reading this as Verizon being extremely conservative in what they are promising today, knowing one day it will do better consistently.
Verizon has a plan on the drawing board to deploy 1.000 MEC servers and bring mobile latency often to 10-20 ms. Apparently, that won't happen quickly. Verizon is severely capex limited. Its millimeter wave 5G is the fastest in the world, often delivering a true gigabit to users. But it only seems to reach ~1% of the country.
The engineers are ready to expand that rapidly if given the money. That's the right move for Verizon. T-Mobile by yearend will have 100-400 Mbps midband 5G to 1/3rd the country, much faster than Verizon. Millions are switching to TMO. Vestberg made a mistake holding back on spending to keep the stock price up.
(Both Verizon and AT&T need to increase capex by ~$10B over the next few years or will be far behind TMO. That's the right move for the company and the net present value of future earnings. Unfortunately, the current belief in the market is that higher capex will clobber stock prices. Because people believe that, the stock price falls despite the underlying earnings value.)
For 80%+ of mobile usage, there's almost no difference between 10 ms and 40 ms. More than half is video. A 4 Mbps video stream looks exactly the same whatever the latency so long as the overall speed is well above 4 Mbps. (4G average speeds are above 40 Mbps on Verizon and most other networks with good backhaul.)
Lower latency, if Verizon actually delivers it, is less of an advantage for gamers than generally assumed. In the most popular games, you are competing against opponents on completely different networks, not on your server. The time to get to the opponents is much greater than the local latency. But gamers look for every advantage. Verizon says privately it expects services like Google Stadia to sign up.
The only application that really does require 10-15 ms latency is virtual reality. Many people get nauseous if latency is above 20 ms. Some have a problem down to 10 ms. Verizon has done massive promotion of the VR possibilities. I spent time last year with the VR people; nearly all believe 10 ms is imminent. I had to disappoint them.
China is building a massive Edge network to connect a billion users. 2020 is mostly a testing year, but by 2022-2023 it will outclass anything in the world. Rakuten in Japan and Jio in India have similar plans.
Unless Verizon, DT, and others speed up the build, the West will be far behind Asia.