Supported Platforms
Related Documentation
Broadband History
Residential broadband services developed using a mainly ATM-based infrastructure and early Internet access required that each subscriber access the network using a dial-up modem to connect from a PC to a Remote Access Server (RAS), or bank of servers, which was connected directly to the Internet. Point-to-Point Protocol (PPP), originally defined by the IETF in RFC 1661, was already in use on leased lines. It was well suited for use on the existing ATM infrastructure and enabled operators to better manage subscriber connections by providing authentication and accounting, along with a level of protocol flexibility due to it being connection-oriented and enabling service providers to customize it to their needs. The use of the PPP model, however, required special software (including the PPP protocol stack) be installed on each PC to communicate within the PPP network. After establishing a connection to the Internet, the subscriber logged in using a PPP user identifier provided by the service provider.
This always on model quickly evolved in several ways. Dedicated broadband access such as DSL replaced dial-up service, replacing the dial-up modem with a DSL modem. Dial-up remote access servers were replaced by the Broadband Remote Access Server (B-RAS) and residential gateways were introduced to allow multiple PCs from one site to connect to the broadband network. Residential gateways have since evolved to provide a wide range of functions including firewall and wireless (802.1b/g/n wi-fi) connectivity. The residential gateway also became the termination point for the PPP connection, eliminating the need for the installation of special PC software.
These new broadband networks were built based on the following two key assumptions:
- Only a small percentage of subscribers were expected to be using network bandwidth at any given time and, even if many subscribers logged in to the network concurrently, few subscribers were likely to enter data at the exact same time.
- Traffic was TCP-based and not real-time. If a packet was lost due to network congestion, TCP detected the loss and retransmitted the packets.
Based on these assumptions, operators over-subscribed the network, enabling more subscribers than a limited amount of bandwidth can support if all subscribers were to access the network simultaneously. For example, if 50 subscribers were to sign up for service that required bandwidth of 1 Mbps for each subscriber, the network did not necessarily need to support a full 50 Mbps of throughput. Instead, operators designed the network to support much lower traffic volumes, expecting maximum traffic flow for all subscribers to occur rarely, if ever. For example, a 50:1 over-subscription needed to support only 1 Mbps of bandwidth. Bandwidth requirements have changed significantly over the years and this method of access is becoming more difficult to maintain.
The basic broadband architecture was initially defined by DSL Forum TR-025 (November 1999). This specification assumed only one service was provided to subscribers—Internet Access (or data). DSL Forum TR-059 (September 2003) introduced quality of service (QoS) to allow broadband networks to deliver voice over IP (VoIP) in addition to data. Because VoIP is a small percentage of overall network traffic, its introduction has not significantly altered the broadband delivery landscape. It is also worth noting that these original standards specified ATM as the Layer 2 protocol on the broadband network.
