5G has gotten a bunch of press lately, and interest is high. This I found to be true when attending the “What You’ll Wish You Knew About 5G
” session at last month’s Enterprise Connect -- the big room was full with people who wanted to learn about this topic. Since then, I’ve been thinking about the major theme: 5G will replace Wi-Fi. I’m not so sure about that premise, and as such have assembled a list of thinking points.
Are you concerned about the security of 5G products, none of which are from U.S. vendors? You have a choice of products from either Asian or European companies. Are there backdoors in any of the products that would allow a foreign party to eavesdrop on your electronic communications? Does this make 5G unsuitable for sensitive or financial data? There are a lot of unanswered questions around the security of 5G hardware.
The 5G network is an extension of the public Internet and we’ll be using VPN encryption to protect our communications. This could create some confusing situations in which someone would be connecting to a corporate resource over a 5G network from within a company building. The 5G nodes wouldn’t fall under the organization’s security umbrella, even though they’re inside the building.
2. Quality of Service
It’s well known that QoS isn’t possible over the general Internet. There are simply too many competing interests -- i.e., packet flows -- to allow QoS to work.
The path from the 5G access points (APs) into the enterprise will be much longer than with 802.11ax-based Wi-Fi installations, increasing latency and jitter. The packets will have to go across the carrier’s network, then get routed to one of the carrier’s Internet exchange points, where they would enter the enterprise network. If the enterprise has a different Internet carrier, the packets would potentially have another network to cross. Latency, jitter, and packet loss would be more prevalent because the longer path would mean more congestion points.
I predict that the carrier would be quite happy to provide a price quote to create a local handoff to reduce the path length. This type of connectivity would assure a tight customer-vendor tie-in and would make it difficult for an alternative carrier to take over that customer.
You’d need a private 802.11ax network if you need QoS for time-sensitive traffic like voice, interactive video, or sensitive business applications. So you’d be back to supporting a Wi-Fi network, which was one of the things that 5G was supposed to replace.
Then there’s the issue of who is responsible for correcting wireless network problems. When (not if) there’s a problem, how is it diagnosed and corrected? For example, we just finished a network assessment in which more than 80 wireless APs were improperly installed or improperly configured (or both) and weren’t functioning correctly. Would your 5G vendor be able to promptly resolve problems that occur in an area where it thinks it has good coverage? Past experience suggests that you would likely need to nag the vendor to get some 5G problems fixed, just as if you had outsourced the installation and management of your Wi-Fi network.
4. Monitoring and Management
When things aren’t working well, how will you diagnose the problem? Your user community will want you to fix it. Knowing what’s broken will depend on the 5G carriers letting you have network monitoring access or at least visibility into their network operations. We found in a recent wireless assessment that the APs in key locations were overloaded. The APs were logging messages about radio overload and we observed client counts that were significantly greater than the vendor’s recommendations. Will our only visibility into the 5G network’s operation be through the number of client complaints? I certainly hope not, though past experience suggests that will be the case.
We’ll probably have to monitor 5G networks in much the same way we monitor connectivity to cloud applications -- using end-to-end application test probes like AppNeta, PerfSONAR, or SmokePing. The problem with these tools is that they report on the secondary effects of overload or misconfiguration, like high jitter, high latency, or high packet loss. While secondary effect statistics can indicate a problem, they don’t provide enough information to identify the cause. Since the wireless infrastructure is owned by a carrier, it’s not likely that you could do much about it anyway.
The RF spectrum is similar for 5Ghz frequencies. However, the millimeter wave spectrum may require closer spacing of APs, due to the reduced propagation, especially indoors. Some forecasts suggest that carriers will need up to 10 to 100 times as many radios for 5G as they have for today’s Wi-Fi systems. How will this sit with a cost-conscious carrier?
I’m also curious to learn who is going to provide the power, conduit space, and cooling for all the 5G support equipment. Sure, there are upcoming technologies like Passive Optical Network (see "GPON vs. Gigabit Ethernet in Campus Networking
") that promise to eliminate a lot of the electronics. These technologies don’t eliminate everything, so we need to be thinking about how the physical plant is going to support 5G and who pays for that infrastructure.
5G carriers are running a business. And that business is to provide network connectivity to the most people at a competitive price. That is a different business than an enterprise that makes and sells widgets. An enterprise can make sure that it has coverage in all the areas that are important to it. A carrier may opt for “good enough” coverage, which may not be sufficient for an enterprise’s needs.
I foresee carriers instituting “installation tuning charges” to get enterprises to pay for providing the necessary level of coverage.
Using 5G for an enterprise network looks a lot like outsourcing the wireless deployment to the carrier. There’ll be charges for the number of APs, perhaps reverse charges for physical infrastructure provided by the enterprise, and charges for each endpoint on the network. This isn’t much different than outsourcing network operations to a managed service provider.
I’m not sure how this will work for some organizations, like universities, where there may be short-duration guests who need service. How will the costs be handled when someone roams either direction between a carrier-owned 5G AP and an enterprise-owned Wi-Fi AP?
I don’t pretend to know all the things that 5G is going to bring or how it’s going to turn out. The above items are some of the factors that I thought about after attending the 5G session at EC19. We have time for them to shake out -- at this point, not many devices support either 802.11ax or 5G -- and we’ll just have to see how the 5G universe develops.