August 1, 2008
Chasing Verizon FiOS
The Race Is On
By Victor Blake
Verizon's FiOS high-speed data offerings are a threat to cable operators. When compared apples to apples (using similar over-subscription ratios) the threat looms large indeed.
But this competition is less a sprint than a marathon. Winning the race depends not only superior throughput today, but also on the correct, long-term alignment of technology, demand and costs. Verizon's current advantage is its use of broadband passive optical network (BPON) technology, but when compared with combinations of DOCSIS and other types of PON, it also represents its Achilles heel.
BPON vs. DOCSIS
Most FiOS deployments have been based on BPON (622 Mbps/155 Mbps), which can offer nearly 20 Mbps/5 Mbps for each of 32 subscribers on a PON. That's the equivalent of half of a 256-QAM/16-QAM (quadrature amplitude modulation) pair shared across hundreds of DOCSIS subs.
Using the same oversubscription ratio as the cable industry does for a 15 Mbps tier on DOCSIS 1.1 256-QAM channel, Verizon could offer a 233 Mbps downstream product. (See Table 1). A very aggressive cable operator might have 75 subscribers on one such DOCSIS 1.1 serving group. The best case - and imaginary - scenario is a 32-sub serving group. (See Table 2.)
TABLE 1: Oversubscription ratio and current product
TABLE 2: Verizon FiOS offering using cable's oversubscribed computation
Using DOCSIS 3.0 at 160 Mbps with four downstream channels, operators could offer a comparable product, but only if it were at 15 Mbps with no more than 10 subscribers per serving group. That's four 6 MHz downstream channels for every 10 subscribers.
But what about the market? If consumers demand less elasticity in the bandwidth offering (a less oversubscribed service), providers will need to offer it. As Internet applications consume more bandwidth and the home gains un-attended devices, the increasing bandwidth demands and service compromises of high peak-to-mean ratios (elasticity) will become less tolerable.
One can build a less oversubscribed service with DOCSIS 3.0. Table 3 shows a fictional configuration to provide a 15 Mbps dedicated bandwidth service based on DOCSIS equivalent to a 15 Mbps PON-based service. At a comparable service level, DOCSIS uses more fiber. Its hardware costs more per Mbps than any PON technology. Verizon has proven how costly it is to build fiber to the home (FTTH) but to the node, DOCSIS 3.0 is some 320 percent costlier than FiOS. (See Table 4.)
TABLE 3: Fictional cable offering using Verizon's non-oversubscribed computation
TABLE 4: Comparing construction costs (Same advertised 15 Mbps DS basis for DOCSIS and FiOS as in Table 3)
BPON, Gigabit (G) PON, and other PON technologies do not typically share access bandwidth until the cumulative total of provisioned bandwidth exceeds the number of subscribers on the PON. Even if we assume the full PON with 32 subscribers and a 1:32 split ratio, they can offer up to 19.4 Mbps, with no oversubcsription.
If Verizon were to oversubscribe, their marketers could say that FiOS is 233 Mbps. That's not all. Verizon is just beginning deployments of GPON, at 2.488 Gbps/622 Mbps. At a recent ITU-T/IEEE workshop, Verizon suggested that its next generation networks would evolve to 10 Gbps/2.5 Gbps, and later to 40 Gbps/10 Gbps PONs.
The competitive threat for upstream bandwidth is a more imminent concern. Even with DOCSIS 3.0, upstream is limited to 40 Mbps of shared bandwidth or 12.8 MHz (optimistically, at 16-QAM).
DOCSIS 3.0 will allow cable operators to market competitive products with FiOS up to 120 Mbps. That certainly leaves room for growth beyond current 10 Mbps-to-20 Mbps residential products. But in the long term, the existing distinctions between the performance capabilities of the DOCSIS and PON networks will become more evident.
From a technology perspective, the business services and residential PON products are all based on the same technology. But not all PONS are created equal. The cable industry is opting for Ethernet (E) PON.
Some cable operators already are using EPON to serve business customers with data rates up to 1 Gbps/1 Gbps. Available dual-speed EPONs can offer 2.5 Gbps/1 Gbps on existing or new deployments. As with GPON, faster speeds lie ahead.
Initiated in March 2006, the IEEE 802.3av 10 Gbps Ethernet PON study group has achieved progress. Chipsets are expected to come to market in late 2008 and early 2009 offering 10 Gbps EPON, far ahead of the telcos' ITU-T GPON efforts. Moreover, next generation EPON will be 100 percent backward compatible with existing EPON. Although there are obviously underlying physical layer (PHY) improvements and some very minor service distinctions, the change will be transparent to operators.
Competitive advantages with EPON are significant. They include lower capital equipment costs, more stable and reliable technology and products, and a lower cost operational model. Cable operators will benefit from innovation, scale and quality by joining with the cable and telco communities in Japan, Korea, India, China and Europe that have already deployed more than 12 million subscriber lines, a number that already has turned EPON into a commodity.
Operationally, cable operators can build an even more competitive edge through the adoption of DOCSIS over EPON technology to extend operations support system (OSS) support to high-end Ethernet services for commercial customers. With a combination of Metro Ethernet Forum on DOCSIS and DOCSIS over EPON, cable's existing OSS can sharpen the competitive edge against the competition.
Strategy and timing
Timing is everything. If the performance advantages are not really necessary now, then cable operator decisions to delay the spending for FTTH will retrospectively appear wise. The risk is that designing, developing and deploying new technologies, products, and services takes time.
If cable operators wait too long, they could lose market share. Object lessons are close at hand. Many dialup operators failed to time their transition to broadband. The ongoing landline-to-cellular telephone service transition is another example.
If timing is critical, strategy is key. Cable operators should support existing product and service offerings; operate both DOCSIS and PON subscribers on a single OSS/BSS, focus on capex and opex, and minimize and technology and economic risks.
More specifically, cable operators should strive for a single OSS for residential and business and use of the same outside plant (optical distribution network or ODN) for both business and services.
DOCSIS over EPON offers all these benefits in a simple technical solution. It is not just "a response" to FiOS; it is a well-thought-out and planned strategy that pre-dates Verizon's FiOS product offering. The challenge is to get the timing right to minimize costs and maximize opportunities.
EPON was designed as a low cost commodity technology. Like the other IEEE 802.x Ethernet standards, the intellectual property behind EPON is royalty free and based on established and proven technologies that can be manufactured in mass at low cost.
Even as Verizon expands its GPON rollout, its prices have not dramatically improved. It is the only large-scale operator in the world that has deployed GPON beyond field trials. At this time, it has only two primary suppliers: Alcatel-Lucent and Motorola. A number of formerly GPON-only vendors are working on new EPON products. Notably, Alcatel-Lucent and Nokia Siemens are active in the IEEE P802.3av 10 Gbps Ethernet PON Task Force.
EPON's high volume translates into opex savings. Another sign of its maturity is the wide variety of EPON lab and field test equipment from numerous vendors. There are also mature product and training technology programs and literature available.
EPON technology is also more stable than the competing GPON technology. While both the GPON recommendations continue to undergo changes, EPON has remained fundamentally unchanged.
The cable operator community has distinguished itself as a strong proponent of IEEE and Internet Engineering Task Force industry standard technologies. Cable operator adoption of Internet protocol (IP)/Internet/Ethernet architecture, for instance, was much more rapid and thorough than its telco competitors. With voice service, cable operators adopted voice over IP (VoIP) to the customer premises long before telcos.
EPON will be a similar challenge for cable. There is no technology risk as to whether it will work. Like VoIP, it is a cost-effective solution than can be integrated with existing IP service architecture and even the fiber distribution plant for 1,550 nm forward video broadcast. While there are some small challenges for two-way video return, the bulk of the technical challenge is in operations, administration, maintenance and provisioning (OAMP).
IEEE 802.3ah and 802.3av are Ethernet protocols just like 802.3z (GigE). They are access protocols that lack service provider-specific OAMP. The telcos are building a service overlay standard for GPON that they call FSAN. The cable community must build an overlay for EPON. DOCSIS is the natural candidate to hide the new EPON PHY/media access control (MAC) layers.
As in previous cases, cable operators can work together to develop a common service interface for EPON that meets specific needs. DOCSIS over EPON is this effort.
It is a competitive advantage that cable operators already have in DOCSIS a strong and mature service overlay technology. Within the DOCSIS family, there is imbedded, best-effort, high-speed data service and voice with PacketCable 1.5. With PacketCable 2.0 and PacketCable Multimedia (PCMM), cable operators can integrate any additional applications with service agnostic quality of service (QoS).
DOCSIS is vertically integrated. It includes the service overlay layers, the MAC, and the various PHYs. Though it was originally organized as a vertically integrated protocol family, we can identify open systems interconnection (OSI)-like boundaries that separate the MAC from the PHY and the service control from the MAC. These layers have been peeled away further as we have added new PHYs such as synchronous code division multiple access (SCDMA) in 2.0, and later the MULPI overlays in 3.0.
The one cross-layer technology is the service identifier (SID). It is no accident that EPON uses a logical link identifier (LLID) as its single, cross-layer service system, just as the SID in DOCSIS. The LLID can do everything a SID can do. Just like SIDs, LLIDs are deterministic and strictly scheduled. They are not probabilistic (like 802.1p and other QoS queuing schemes).
Operationally, cable operators can build an even more competitive edge through the adoption of DOCSIS over EPON technology to extend OSS support to include high-end Ethernet services for business customers. With a combination of MEF on DOCSIS and DOCSIS over EPON, the existing OSS can further the competitive edge for cable competition with FiOS.
It is possible to envision a continuous transition. Because all traffic going to and from an optical network unit (ONU) must pass through the optical line terminal (OLT), EPON technology can hide behind the OLT, acting like a cable modem termination system (CMTS). The complexities of the service interfaces in CMCI, for example, can be hidden by the OLT. An EPON system running DOCSIS over EPON can make an ONU act like a cable modem, using CMCI, without a single code change to existing and shipping ONUs!
DOCSIS currently supports Layer 2 virtual circuits through the L2VPN protocol in DOCSIS 1.1 and 2.0. In DOCSIS 3.0, these same services and additional functionality are available in the Business Services over DOCSIS (BSoD) specification. L2VPN and BSoD offer a point-to-point (P2P) Layer 2 virtual private network (VPN) from the modem to the CMTS. Thereafter, service providers can use off the shelf protocols to extend the L2VPN to its far end destination. Recognizing the limitation of this access-only transport, cable operators are actively working on a CableLabs effort to incorporate MEF OAMP compatibility into BSoD.
Since many cable operators already use EPON solutions (some MEF compliant), it would be preferable to have a single solution that works across both the DOCSIS RF and DOCSIS over EPON platforms. By incorporating MEF into DOCSIS (on existing CMTSs) and working to do so for DOCSIS over EPON in the future, cable can create an end-to-end system to provide seamless P2P and point-to-multipoint (P2MP) Ethernet services as outlined by the MEF.
Ethernet transport services are not just for "commercial." Ethernet could well be the fifth play in residential. Economics continue to drive more companies and employers to work at home or remotely. Challenges with software, configurations, and do-it-yourself IP-VPN solutions leave many companies looking for a simpler solution. MEF services to the home could be a significant market opportunity to add on a new service and create a new revenue stream in residential dwelling units, but funded by business service opportunities.
Each cable operator's unique and specific transition strategy will likely vary by market. The most significant factors influencing an FTTH decision will include variations in constructions costs, the competitive market, and most importantly the timing of capital expenses.
A combination of residential DOCSIS 3.0 and commercial EPON services in can compete with FiOS. DOCSIS over EPON will offer a seamless transition from a vertically integrated DOCSIS over RF to a future with DOCSIS over RF/EPON (and even other transports).
With DOCSIS over EPON, cable operators can time their introduction of PON services according to when it makes the most economic sense. In greenfield deployments, that might be now. For overbuilds, it may vary depending on the ARPU and expenses. Combined with DOCSIS, EPON isn't just competitive - it is superior. The reduced operational costs of a single seamless OSS transition will translate into a lower cost basis and greater profitability for cable operators.
Victor Blake is an independent consultant. Reach him at firstname.lastname@example.org. Drawn from a paper presented at the SCTE Cable-Tec Expo.