Posted by: Tom Nolle
100 Gigabit Ethernet, Alcatel-Lucent, edge networks, FP3, network monetization, routers and switches
Alcatel-Lucent provided its own ballyhoo with the announcement that the company had promised would make the Internet faster. I’m not big on ballyhoo, and I have to admit that I have mixed feelings on the Alcatel-Lucent announcement. I want to be fair, so I want to start with the perspective I bring to the issue of “advancing the performance of the Internet.”
I’m a strategy analyst—someone who surveys and models markets. My goal isn’t to find out what people want, but rather to find out what’s going to happen and what’s not. People want gigabits for nothing, but that’s not going to happen. We could give people faster broadband today in a technical sense, but the decision would fail for financial reasons. You could argue that anything that reduces the cost of the much-for-nothing goal is an advance, but it’s doubtful that any single development could create a cost revolution.
That is the dilemma Alcatel-Lucent’s announcement poses. The company has made an impressive technical stride that I’m not confident is a significant market stride.
Alcatel-Lucent’s FP3 chip is a quite revolutionary achievement in special-purpose network semiconductor design and fabrication. This chip, says Alcatel-Lucent, is faster (by 4x), smarter and greener (50% less power per bit) than ever before, and the company says that it will accelerate the adoption of 100 Gigabit Ethernet from edge to core. Certainly, the chip can support multiple 100G interfaces or a future 400G interface, but the question of capacity in my view is less one of technical performance than of financial performance.
Operators we survey are watching 100G Ethernet to be sure, for a time when it would be economically justified. They don’t think they’re at that point. The problem they have is that revenue per bit is already plummeting. A four-times-faster chip would presumably need a network with four-times-higher capacity. Our model says that consumers will pay 17% more for 4x speed; operators are estimating 20% more on the average. The uptake for premium speed tiers is low, and FCC data shows that broadband users in the U.S. cluster at the low-cost end of the service range. So how does making the network capable of higher performance change things fundamentally?
Something for nothing (or next-to-nothing) may be appealing, but it can’t be delivered. The thing is, people are willing to pay for stuff; it’s just that they’re not willing to pay for Internet bandwidth. Apple’s success with apps demonstrates that people will shell out millions for just the convenience of using an app for what they could get from a website. If there’s a revolution in the market, it has to come from allowing the people who build the networks to participate in this higher-level part of the food chain. They want to do that; they’ve told me that explicitly in surveys for four years now. They want to add services to their networks, to add a service layer to their network layer.
Which brings us to “smarter,” the third claim for the FP3. What exactly does that mean?
Alcatel-Lucent says that “smarter” means fully programmable delivery of personalized services and content, and massive IPv4/IPv6 scale for the future. But unless we believe in 100G to the user, the access network will have to be able to do all of that or the user will never see the service, and the FP3 won’t be out there.
We all know that supporting most of the “smart” things is likely an edge role. The power of the Internet was to avoid being aware of individual users or flows deeper inside the network; it doesn’t scale. And what exactly is the FP3 programmable to do? Yes, the number of VPLS and virtual private routing network (VPRN) instances is doubled, along with the number of queues and (almost) the number of routing table entries. The question is how exactly this creates monetization—the increased revenue-per-bit that operators need if they’re to punch capacity up by 400%.
Alcatel-Lucent did offer a couple of ideas on services—one on the general evolution of the service experience and one on the video distribution process. I agree with the points in both; what I still have a problem with is what role the FP3 plays beyond moving the bits around.
I doubt that Alcatel-Lucent proposes to add customer or service-flow awareness to deeper aggregation products that have the traffic scale to justify 100G. It doesn’t scale. Is Alcatel-Lucent proposing some intermediate “not-aware-but-sentient” role for the network? I’d love to hear about that, but I didn’t hear it in this announcement.
It’s possible that the FP3 could play a role in binding the services of the future to the network of the future, but Alcatel-Lucent doesn’t say that. It’s possible that Alcatel-Lucent intends to meld its Application Enablement and Open API themes downward into the network and create a multi-layer profit partnership, but it doesn’t say that either.
The FP3 is specifically faster, specifically greener and un-specific about how it’s smarter. And it’s smarts that will revolutionize the Internet; it’s smarts, not just bits, that generate bucks. For a company that has gained router market share because of its success in the mobile and content service layers, it’s disappointing that Alcatel-Lucent would forget a strong technical service-layer tie here. It would be more disappointing if there isn’t one.
Without monetization in a service sense, traffic can’t be profitable even at current prices, and the FP3 presumes a 400% traffic gain. Long before we reached that, the current market model would collapse into usage-based pricing, which would limit traffic growth and also the growth of the Internet. We have to create a healthy ecosystem here, and the FP3 picked up three credible points about that ecosystem: speed, smarts and power efficiency. It validated two of the three.
So that’s my dilemma. I think the engineering is impressive—in fact, very impressive. I think it could reduce the cost of high-capacity devices, and just the fact that Alcatel-Lucent announced it may suggest that the company is planning to go higher on the router capacity tree. I just don’t think Alcatel-Lucent has proved that it revolutionizes the Internet, because nothing is going to do that except something that revolutionizes the Internet business model. The capacity play it has made for the FP3 is dangerously close to following Cisco down into the “bandwidth at any cost” abyss that I warned about when Cisco announced its ASR 900 enhancements.
The FP3 shouldn’t have been about speeds and feeds, but about dollars and cents. The “smarts” point of the FP3 launch is the one that had to be the strongest; instead, it was the weakest. Might Alcatel-Lucent plan to correct that down the line? Perhaps, in which case, I’ll take another look when they announce it. For now, this is a strong evolution, but it’s not a revolution.