Last week, Google Fiber made another big pronouncement that it has added 34 more cities in nine U.S. metro areas to its list of potential fiberhoods. This, of course, generated a massive amount of publicity for Google Fiber in the U.S. and globally. In the meantime, across the Atlantic, the FTTH Council Europe held its annual conference in Stockholm, Sweden, where we learned that the Google Fiber network is not Ethernet FTTH, but (drum roll, please)…GPON!
When Google Fiber first declared its intentions several years ago to build a Gigabit-capable fiber access network in the city that made the strongest case for itself (i.e., Kansas City, USA), the assumption from the beginning was that Google Fiber would leverage an active, point-to-point Ethernet FTTH access network technology (and architecture) to deliver its symmetric, 1 Gbps connection to each household.In fact, Google Fiber extensively tested active/P2P Ethernet FTTH systems from multiple vendors (e.g., Cisco, Alcatel-Lucent and Zhone). However, over the past several years – not coincidentally, as the Google Fiber network was actually built out in Kansas City – there have been assertions from networking professionals and others (including knowledgeable Google Fiber customers) that the connections are in fact GPON.To be clear, the exception to this is the former iProvo network in Utah, which utilizes the legacy active/P2P Ethernet network built years ago under previous ownership, and which Google Fiber bought in April 2013 for $1 (USD).
So,does it really matter that Google Fiber is using GPON instead of active Ethernet?. For Google Fiber customers, the answer is no, as most end users care far less about technology than the service delivered.For network operators and their networking systems suppliers, however, it matters a great deal.The advantage of active/P2P Ethernet is that each end user is provided with a dedicated Gigabit Ethernet port at the operator’s central office, which can be 10/100 or 1000 Mbps (1 Gbps), in contrast to a PON system, where each optical line terminal (OLT) port serves multiple end users by virtual splits (hence, the point-to-multipoint PON architecture). Split ratios are variable, but are ideally in multiples of eight (e.g., 1:16, 1:32).
Active/P2P Ethernet is superior in terms of delivering dedicated, symmetric bandwidth, but the reality is that very few end users are going to utilize a dedicated Gigabit Ethernet port, especially on a “24/7/365” basis. Simply put, dedicating a GigE port to each user is not only extremely expensive, but vastly underutilized in terms of traffic throughput, even for heavy users.
Google Fiber’s choice of GPON to deliver 1 Gbps symmetric services makes sense on several levels, the most important being capital costs. In addition to the central office cost of 1 GigE port (on an Ethernet switch or OLT) for each end user, the expense of “stringing up” hundreds (let alone thousands) of dedicated optical fibers on the aerial plant in Kansas City, for example, is dramatically higher compared to a GPON network, where each OLT port serves multiple end users (again, 16, 32, etc.), and the PON network has a much smaller fiber density in the field.
In closing, it is important to remember that Google Fiber is not the only Gigabit service provider to leverage GPON technology/architecture, or even the first. In the United States, Chattanooga, Tennessee’s Electric Power Board (EPB) was the first to offer symmetric Gigabit services (to a sizeable serving area, that is), and in China, Hong Kong Broadband Network (HKBN) has offered the world’s cheapest ultra-broadband, Gigabit connection to its primarily residential end users at approximately $29/month.Both of these operators have offered these services for at least four years, well ahead of Google Fiber (incidentally, both are using Alcatel-Lucent’s GPON systems).As more operators begin to offer Gigabit connections, they will undoubtedly leverage GPON systems and architectures to deliver these ultra-broadband services.