Evolution or Revolution?
In the communications industry, we often refer to unification as being inevitable. We want more services across a common set of platforms and technologies in the interest of scale and sound business operations. But the fiber access market continues to challenge broadband providers with many different options and discrete products tailored to regional sets of requirements. At least, this has been the case up until now.
In 2011, we expect a new generation of Optical Line Terminals (OLTs) to hit the market. They will be more capable and far more flexible. And they will successfully concur a global market. The secret recipe? Unification.
There are a number of reasons why there are so many access technologies competing for market attraction: demographics differ, legacy investments in copper loops vary in quality and length, regulatory policies often make one type of solution more favorable than others, etc. In addition, increased competition drives broadband service providers to carefully consider the options to strike the right balance between cost, performance, reach, and time-to-market.
The cost of deploying fiber to the premises is two-to-five times the cost of enhancements to the existing plant. This makes upgrades to copper and cable very attractive, such as fiber-powered DSL (Fiber-to-the-Node, VDSL2+fiber) and DOCSIS 3-channel bonding. The picture is fragmented, and there are good reasons to believe the market will continue to stay in a mood of divergence.
Successful Local Players
The FTTx market is supported by a broad set of access vendors. The hot spot is in Asia, where local players in Japan (Fujitsu, Hitachi, Mitsubishi, NEC, Sumitomo), Korea (Dasan, Corecess) and China (Huawei, ZTE, Fiberhome) have been very successful in supporting their respective local markets. Products tend to be optimized to the set of requirements in each country. And while standards continue to evolve, dominating service providers have defined de facto usage of technology for their respective markets. The ways broadband networks are managed, operated, and maintained differ from provider to provider.
Access vendors need their systems to be flexible enough to adapt to local market conditions. Some of the global system vendors have learned this the hard way when their products could not support the required model of operations, in practice locking them out of a strategic market. Other vendors owe the luxury of incumbency; once a system is “in-designed” it becomes “approved” by the service provider. This puts vendors with running contracts in a strategic position where they may advise new features that are planned for in their future releases of their operating system, and are well supported by the underlying hardware in next generation platforms. Competitors not in this position struggle to catch up and get their systems approved.
Inflection Point
There are a couple of reasons why a new generation of OLTs and access switches are demanded. The 10G versions of EPON and GPON are developed to support more bandwidth down to MDUs connecting multi-tenant buildings. These technologies require more bandwidth than can be processed by current designs. Increased user penetration and service uptakes push for higher port and service densities to obtain lower cost per user. In addition, cost-effective packet processing and traffic management silicon are now available to build more capable access devices. This helps scale service provisioning and policy enforcement to the access domain. All together, a new generation of OLTs may push the market in several potential directions.
In this inflection point, there will be vendors hitting the right mark in time-to-market and right-sized products. What are the characteristics of these systems and how do they differ from current products? Figure 1 provides an overview.

Figure 1. A new generation of OLTs and access switches has more bandwidth and is more advanced from current products.
When looking back on other inflection points in the broadband industry, such as the move from RAS to BRAS, from BRAS to Carrier Ethernet Edge Routers and the introduction of IPTV, one thing stands out: the most agile and quickly moving vendors will benefit the most. Products based on programmable silicon with the right capacity help these vendors to add and customize features to align their roadmap closely with key service provider customers.
Technical Requirements for the Real World
The access switch or OLT needs to support a range of First-Mile-specific standards (point-to-multipoint PON or point-to-point Ethernet). These are implemented in PON controllers, ASSPs available from a range of vendors. To support the unified trend, there are now PON controllers that share pin-out and features between GPON and EPON.
In addition to access link specific features, there are quite a number of features that are common for any type of access, as illustrated in Figure 2. This calls for a split architecture of the OLT/access switch, where user session services are designed around a programmable device, such as Programmable Ethernet Switches for user plane implementation. To effectively unify the design and quickly go to market with a proper range of products, it is vital to reuse development efforts as far as possible. Being strict regarding where functions reside helps facilitate this.

Figure 2. The set of requirements for next-generation OLTs can logically be divided by access link oriented features and subscriber session services and network services, implemented by PON controllers.
Vendors looking to quickly introduce different types of systems can further leverage the unified approach if the data planes are kept intact between pizza boxes and chassis-based systems. In chassis, the traffic will pass through a backplane, which is typically a bottleneck resource requiring careful traffic management and queuing intelligence. This adds to the complexity of the design not present in pizza boxes. It is therefore tempting to optimize each system individually, resulting in higher performance for the invested hardware dollar.
Others, however, take the unified approach to the full extent and reutilize everything in the design. This significantly reduces system tests and streamlines software maintenance. The vendor benefits from having a more common development environment, and can get new features to market more quickly than its competitors. In times of intense feature competition, this can prove to be a strategic winning concept.
Figure 3 provides a high-level overview of different access systems that can be developed based on a unified data plane. As an example, Accton provides an Ethernet access pizza box featuring as much as 48 GE client ports and four 10GE ports for uplinks, all in a 1U rack mountable unit. The level of performance and integration is made possible through a new generation of programmable Ethernet switches with integrated traffic management as well as GE, 10GE MACs, and PHYs.

Figure 3. Different types of OLTs can be designed using a unified development approach.
Taking the design to a chassis-based system is straightforward. In compact chassis designs, the programmable Ethernet switches can manage the traffic flows and switch traffic between the line cards using inter-traffic management communication methods to manage queues across the system. Larger systems utilize a 48-64 port KR/XAUI switch to interconnect the line cards. Recent Ethernet enhancements like priority flow control (802.1Qbb) and congestion notification (802.1Qau) add methods to communicate the status between subsystems in this design. An example of such OLT/access switch design is illustrated in Figure 4.

Figure 4. Example of a unified chassis-based OLT/access switch.
The trend towards unified designs is not limited to the fixed broadband world. Several mobile operators have made a policy decision to utilize fiber to the base stations to the greatest extent possible. What traditionally is called mobile backhaul may become a different service in a unified fiber access infrastructure. This has several implications on the data planes of the access nodes. The hardware and software need to support Precision Time Protocol (PTP) and Synchronous Ethernet. The traffic flows across the system need to take additional types of user and control traffic into consideration for careful design to meet requirements on packet loss, delay and delay variations. Programmable Ethernet switches are prepared for this evolution, paving the way for system unification. The rate of acceptance of this will determine the pace of convergence in the networks. It is likely to take some time, but with China being a large market and for pure Ethernet-based mobile backhaul, we may see things happen sooner than expected.
Unification is a strategic design move by several PON and Ethernet access vendors. It will help them compete more successfully on a global basis. Service providers will benefit from more flexible solutions supporting their long-term requirements. We can expect another technology push to further fuel fiber-based broadband uptake.
Per Lembre is Director of Product Marketing at Xelerated. He has more than 10 years of experience in product management and marketing in the data networking and communications industry. For more information, email per.lembre@xelerated.com or visit www.xelerated.com.
What’s your take on this subject? Leave a comment and get the conversation going.
