In the Net neutrality debate, network engineers believe the Federal Communications Commission's proposal to allow...
Internet fast lanes will harm the "Open Internet." What's more, they also don't think fast lanes will work.
Net neutrality debate: What are fast lanes and will they work?
"I've read some of [FCC chairman] Tom Wheeler's writings and some of the cable industry's writings on this paid prioritization, and they say it won't slow down [other] traffic," said Andrew von Nagy, a network engineer and a technology evangelist for a wireless infrastructure vendor. "From an engineering and architecture perspective, that is flat out wrong. Pushing packets through a link is fundamentally a zero-sum game. If you prioritize some packets, you have to slow others down. The industry's talk about how prioritization won't create slow lanes and will just create fast lanes is just fundamentally wrong."
The public needs to think about the conditions under which Internet fast lanes come into play, he said. They only matter when a network is congested and everyone is suffering, he said. That's theoretically when quality of service (QoS) tiers would take effect and open up a fast lane.
"When there is no congestion, there is no need for paid prioritization," von Nagy said. "This proposal from the FCC [Federal Communications Commission] is incentivizing ISPs to some extent not to invest in their network architecture or to artificially create bandwidth scarcity -- either at the last mile, or to create artificial scarcity elsewhere in the network where they can control the flow of information, such as what we've seen with other paid peering agreements like Netflix and Comcast."
Given that most Internet traffic hops across multiple ISPs from its source to its destination, the concept of Internet fast lanes has some engineers wondering if such a thing can even work.
"It works well on the private [WAN] side, when we have control end to end, but there is not going to be any consistency for actual end users when running through a half dozen providers to get from point A to point B," said Matthew Norwood, a network architect for a systems integrator. "Netflix might have an agreement with some of these national service providers, but what if I have Internet from a smaller provider that doesn't have an agreement with Netflix because they don't have the capacity? It may go through three or four service providers that all treat that traffic differently, and it just becomes chaotic for the end user."
Net neutrality debate: The difference between QoS and fast lanes
Some might argue that ISPs apply QoS all the time to protect their infrastructure and ensure customer satisfaction. From that perspective, why should ISPs be free to offer premium tiers of QoS to those willing to pay for it?
"I worked for an ISP for a long time," said Nick Buraglio, a network engineer at a global research network who has also worked for a large Midwestern university and a rural ISP. "When I was running the show, we did no filtering unless there was an imminent risk of degrading the network, if we were under a denial-of-service attack, or [had] a security issue. I did, from time to time, have to employ some traffic shaping on some apartment complexes when Kazaa came out because two people would just trash a T1 line. So I had to do some really basic traffic shaping."
Fast lanes and paid prioritization is less about network optimization and more about business arrangements, von Nagy said. In networking terms, network optimization is to Internet fast lanes as the OSPF routing protocol is to Border Gateway Protocol (BGP).
"What do you do inside your network for an interior protocol?" he asked. "You might do OSPF [Open Shortest Path First] or EIGRP [Enhanced Interior Gateway Routing Protocol], which is purely optimization of routing in order to provide the most efficient path and highest resiliency and a good user experience. But when you get to the edge of your network, you typically implement BGP -- which is much more focused on policy decision making -- where you have more control over the business arrangements between you and multiple providers."
Beyond Net neutrality
The FCC's reluctance to treat ISPs as utilities will have consequences beyond Net neutrality. Bandwidth caps, for example, are starting to work their way into the broadband market, much like they already proliferate in the mobile data market.
Those caps, and other steps ISPs might take to throttle throughput, could usher in a whole new set of costs and performance problems engineers would have to contend with.
"Telecommuting could dramatically change based on Net neutrality changes," Buraglio said. "There are a lot of people who work in the field I work in who require higher-bandwidth throughput to do their work. I'm on a video conference all day, every day. It uses 4 Mbps, it's HD and it's going all the time. Putting data caps on my connections at home might force me to bump up to a higher [service] tier to do those things.
"They could start blocking ports, or [they could] say, 'You can't use the [corporate] VPN if you are on a consumer-grade connection. You have to upgrade to a business package for that.' And then they could block those ports. Eight or nine years ago [an ISP] blocked port 500, which is the IPsec port, and if you had their consumer-grade DSL line you couldn't connect to the corporate VPN unless you bought their business package, which was three times as expensive."
Net neutrality or not, liberate the last mile
With the FCC debating national policy, what can the individual network engineer do? Not much. But on a more local level, perhaps engineers can make some noise.
"I'm not a fan of big regulation, but if [fast lanes] start happening, people will get so disenfranchised that guerilla networks will start popping up again," Buraglio said. "They will find copper lines somewhere, put their gear in there, start running their own POPs [point of presence], and start buying connectivity from somebody. Then the phone companies are going to get wise and say, 'You can't run over alarm circuits anymore,' and they'll kill those off.
"At the same time, encrypted VPNs will come up and start port-jumping, which are real hard to track, which will force service providers to put deep packet inspection everywhere. Then the municipal fiber networks will start to pop up and trade with carriers by dark fiber or managed wave to somewhere else where they can get an Equinix POP to Hurricane Electric or someone else. And they don't care about Net neutrality. They're just a transit provider."
Individual communities abdicated control to cable companies decades ago when they granted exclusive franchises. Some advocates say cities could take back that power, perhaps taking over the last mile and interconnect to multiples ISPs. Then consumers could get Internet service from any ISP they wanted and create real competition. Some cities have already gone down this road, such as Chattanooga, Tennessee., which used a federal grant to allow its local, publicly owned utility to build a city-wide gigabit fiber network.
ISPs are already lobbying hard against local efforts to create municipal last-mile infrastructure. Plus, such efforts won't be cheap. But if ISPs do start running fast lanes and continue to add other things like broadband data caps, consumers might force the issue.
"My personal feeling is that the last mile needs to be customer-owned and managed by the municipality as a public service," said Dale W. Carder, senior network engineer for the University of Wisconsin-Madison. "For example, I own the sewer lateral pipe from my basement to the middle of the street. A similar model can work for fiber connectivity, where the customer owns the drop from the poll or the nearest splice case," he said. "The municipality manages all of this and aggregates it to a set of common meet-me points where any service provider may interconnect to any customer. Then it's up to the customers to decide who has the service they want and at what price."
To read the first part of this story click, "Network engineers on Net neutrality debate: Protect Open Internet."