Putting Huawei in a (White) Box
5G is the most important development in networking since the 2G mobile network was created in the early ’90s. 2G put wireless communication on a digital foundation, opening the door to massive progress in years to come.
5G will bridge the access network performance gap between wired and wireless. While the Internet’s core and backhaul will always be wired (for the most part), there’s enormous utility in enabling pervasive access without a performance penalty.
We can see this in the way we use Wi-Fi in homes and offices to do things we were once forced to do over Ethernet. This is why there’s so much hype around 5G.
Obstacles in the Path
Apart from pilot projects, the 5G rollout hasn’t really started yet, and won’t get going in a big way for another year. But we’re already seeing a cottage industry developing that seems to be intent on slowing it down if not stopping it altogether.
The anti-5G movement is populated by hipster journalists, technophobes, and commercial interests deeply invested in the status quo. This isn’t surprising given that every leap forward in technology is always contested.
I don’t doubt that hipster cave people were cynical about fire and the wheel. In their own ways, they disrupted social orders built around raw food and walking.
Safety and Security
The same people that insist the earth is flat, the atmosphere is full of toxic chemtrails, vaccines cause autism, and electricity causes cancer are now warning us about a 5G menace to public health. Mainstream media publishers such as Condé Nast are also in the scrum.
Wired magazine had to make wireless critic Susan Crawford rewrite an opinion column that used electro-magnetic hypersensitivity nuts as sources for her attempt to whip up health and safety fears about 5G systems operating in totally different frequencies from the power lines that scare EHS loons. Crawford is easily dismissed, but the security concerns come from far more reputable sources.
According to the CIA, strong evidence suggests Huawei is funded by the Chinese government’s security services. Huawei has a history of intellectual property violations, most of them against Cisco. Huawei’s first products were clones of Cisco routers that used actual Cisco software.
Huawei has also been accused of attempting to violate patents on diamond glass and of building back doors into its equipment. These claims are often wildly overblown; the so-called “back door” was simply a telnet command line server in a home router.
Home router developers (such as myself in a former life) use telnet – or its more secure cousin SSH – for testing and debugging software. We’re supposed to disable this feature before shipping gear to customers, but sometimes we forget.
Telnet and SSH are optional features of every home router on the market, so the only issue here is whether it’s enabled or disabled by default. Unlike a true back door, an open telnet port is easy to find. So no, the so-called Huawei back door is more like an open front door to an edge device that lacks the ability to affect the network core despite what you may have read.
Open telnet ports with default passwords are a significant vulnerability: the Mirai botnet attack exploited this flaw in cheap Chinese video cameras. But ISPs test devices for such vulnerabilities before deployment because such mistakes are not unheard of.
Significantly, Huawei is emerging as a dominant supplier of 5G gear because its products are cheaper and often more functional than competitive products offered by Ericsson and Nokia. Even loyal customers of Ericsson are using some Huawei equipment these days.
Deep financing, a strong focus on advanced features, a willingness to cut corners on intellectual property, and low prices all add up to both happy customers and a distorted market. As long as Huawei doesn’t become too dominant we don’t have all that much to worry about.
Applying Software Discipline to Software-Defined Networks
The main difference between 5G and legacy networks comes down to the role of software. 5G is a virtual network, an overlay of software built on top of traditional network hardware.
This is how we get network slicing, the ability for customers to get precisely the kind of network service they need to get a job done. Virtual networking is the soul of the Internet, the first network that allows applications to run on a wide variety of hardware as if it were (pretty much) all the same.
But virtual networking doesn’t mean that hardware suppliers can instantly inject any software they want into the network at any time. While operators can do this, they test software updates before they deploy them, whether the updates come from the operator’s own teams or from equipment vendors.
A lot of the fear of Huawei builds on misconceptions about software engineering that ignore best practices. Not only do we test before we deploy, we generally have the means to quickly back out changes that prove to be problematic.
Banning Huawei Creates a False Sense of Security
The current controversy over Huawei features ever-changing allegations and the false belief that relying on trusted suppliers is sufficient to ensure that networks operate as desired. Huawei may very well be the worst company in history, infested with spies, and beholden to the Chinese Communist Party.
But that doesn’t mean nobody should do business with them as long as they have the best technology and the best value proposition. It does mean that it’s reasonable to use Huawei equipment to do jobs that are carefully monitored.
If monitoring networks for suspicious activity seems like an onerous burden, ask yourself if you want your personal information to be carried by a network that doesn’t monitor as a matter of course. It’s not like putting a bullet in Huawei’s head is going to instantly resolve all of the Internet’s security nightmares.
Life After Huawei
While the security concerns about Huawei are overblown, a market with one dominant supplier and a couple of shrinking violets is uncomfortable because the dominant player controls standards, capabilities, and the pace of progress. We don’t ever want one company building fences around innovation regardless of how virtuous they may be.
IBM’s dominance of mainframe computing was a bad thing while it lasted because of high prices and a diabolically difficult upgrade process. But in the fullness of time, that dominance led to the creation of minicomputers, personal computers, and the cloud.
Huawei excels at specialized hardware complemented by even more specialized software. But the heart of networking isn’t base stations, switches, and routers, it chips and software.
Several Companies Make their own Routers
Google, Facebook, and Amazon make their own Ethernet switches. Google pioneered this practice a long time ago:
According to [Google Fellow Amin] Vahdat, Google started designing its own gear in 2004, under the aegis of a project called Firehose, and by 2005 or 2006, it had deployed a version of this hardware in at least a handful of data centers. The company not only designed “top-of-rack switches” along the lines of the Pluto Switch that turned up in Iowa. It created massive “cluster switches” that tied the wider network together. It built specialized “controller” software for running all this hardware. It even built its own routing protocol, dubbed Firehose, for efficiently moving data across the network. “We couldn’t buy the hardware we needed to build a network of the size and speed we needed to build,” Vahdat says. “It just didn’t exist.”
This is something that happens when firms with private networks reach a certain size. Facebook has an even more comprehensive approach to data center computing as a whole:
During Facebook ‘s remarkable 11 years from launch to 1.44 billion monthly active users as of the first quarter of this year, the social platform outstripped the capacity of existing networks to carry its traffic. So Facebook had to build four data centers at the heart of a global network, and design its own hardware and software, all to make sure that you don’t have to wait a moment to see that photo of your nephew’s trombone recital.
Facebook turned to its own designs because the service was moving too fast for networking vendors to keep up.
And Amazon has long toyed with the idea of white boxing routers:
The [Amazon] switches would specialize in connecting private data centers and AWS. They’d be built based on open source software and unbranded hardware, with built-in connections to AWS cloud services. AWS uses similar white box switches inside its own data centers. It expects to launch the network switches for customers within the next 18 months, working with white box manufacturers such as Celestica, Edgecore Networks and Delta Networks, according to The Information.
The keys to this design approach are specialized chips made by Broadcom and Qualcomm, generic CPUs, and open-source software.
Open Compute Project
Facebook took the initiative to create an open source project for white box routers in 2011 called the Open Compute Project. Its members now include the most important players in networking.
OCP member AT&T created an “Open Architecture for a Disaggregated Network Operating System (dNOS)” a year and a half ago; last year the Linux Foundation created the DANOS project to host it. Last October, AT&T released cell site backhaul specifications to OCP.
So it may well be the case that Huawei will continue to dominate the market for small cells for the immediate future. But these cells will use very similar technology to the 5G modems in the end user devices that use the 5G network and the network backhaul and network cores will be generic.
Now that We’ve Settled That, How About we get on with 5G?
OCP limits the damage that any one supplier of 5G equipment can do in the worst case. 5G networks are and will be software-intensive. We need to keep our eyes on China just as we remain vigilant about Russia, but Huawei doesn’t deserve the demonization they’re undergoing from our defense establishment, the Russophiles, and the hipster antitrust movement.
I suspect – but cannot prove – that a great deal of the Huawei fear-mongering comes from the same parties that wish to nationalize 5G. They’re relying on technology ignorance on the part of the public and the tech policy blogger communities to advance their cause.
There are issues that warrant special attention in the networking standards bodies (3GPP and IETF in particular) but this is nothing new. If the government can get off its Huawei kick and support OCP we’ll all be better off in the long run.