PBT: Too soon to ask what’s next?
by Kevin Daines
It seems somewhat premature to ask, “What’s next for PBT?” when Provider Backbone Transport (PBT) is only in the initial stages of investigation by carriers and operators. However, PBT is a technology that has the potential to change the telecommunications landscape-and, as such, it is at an important crossroads.
PBT has recently emerged from being considered a relatively obscure and proprietary transport technique to gaining broad support as a future Institute of Electrical and Electronics Engineers (IEEE) 802.1 standard. Known as the IEEE 802.1Qay Provider Backbone Bridging-Traffic Engineering (PBB-TE) project, the nascent technology brings added resiliency and provisioning capabilities to the popular family of international Ethernet standards. In this article, we’ll explore exactly what PBT is, why it is relevant, and what the future holds for this promising, innovative technology.
Since the late 1990s, Ethernet has been successfully deployed in access and aggregation networks, and this trend is increasing as carriers and operators choose Ethernet for its simplicity, performance, and cost-effectiveness. Traditional Ethernet-based metro and access networks typically support a connectionless model for delivering Carrier Ethernet services. While the endpoints of a given service are known, the precise path is left up to the IEEE 802.1w Rapid Spanning Tree Protocol (RSTP) control plane. RSTP has proven sufficient for most topologies with a moderate number of network elements. RSTP automatically prevents loops and broadcast storms and ensures Ethernet services are delivered across a set of Ethernet networking devices.
Figure 1 shows how RSTP is used to prevent loops and resultant traffic storms in the network. The control protocol automatically blocks various links (depicted with an “X”) and monitors the state of the network to respond to topology changes and equipment or link failures. Several customers are interconnected across this RSTP-enabled Ethernet network. However, network and path utilization is suboptimal as some links are disproportionately overloaded or unused.
Many domestic and international carriers are interested in capitalizing on the advantages of Ethernet and have begun to create ever larger packet-based access and aggregation networks to reduce overall network costs and provide more scalable, higher-bandwidth services. Frequently, they want a more cost-effective method of building metro networks that leverage existing investments in their multiprotocol label switching (MPLS) core. Other carriers want to provide Carrier Ethernet business or residential services. In some instances, established and emerging carriers want more scalable and reliable techniques for transport and backhaul of non-Ethernet technologies such as DSLAM aggregation or packet-based 3G/4G mobile services.
PBT proponents believe that carriers are ready for a connection-oriented, native Ethernet approach that provides explicit Ethernet switched paths. As Fig. 2 shows, primary and backup Ethernet switched paths provide redundant interconnectivity and path diversity. Network utilization is increased, and higher-bandwidth services can be delivered.
The network in Fig. 2 is made possible through the use of traffic-engineered Ethernet switched paths, also known as tunnels. The traditional Ethernet RSTP/Multiple Spanning Tree Protocol (MSTP) control plane, source address learning, and unknown address flooding features are disabled for specific, administratively selected VLAN identifiers. In its place, PBT uses a network management system to create static forwarding entries in each transit and edge device.
Within a Layer 2 network domain, each switch learns the MAC addresses of all connected devices. The size and reach of the Layer 2 network is constrained by the scalability and performance of the switches. In addition, customer and service VLANs are limited to 4,094. To overcome these restrictions, PBT leverages the IEEE 802.1ah Provider Backbone Bridging project’s MAC header encapsulation technique to provide for greater network, MAC address, and service scalability. Figure 3 illustrates the MAC header encapsulation technique.
Backbone edge bridges (BEBs) that support PBT perform the MAC header encapsulation function. The destination address of the BEB on the other end of the tunnel is added to the front of the frame along with its own MAC address. The administratively selected VID is inserted as well as an instance tag. The 24-bit instance tag (known as the Instance Service Identifier), with a range from 1 to 16 million helps to overcome the 4,094 VID limitation of service and customer tags. This encapsulated PBT frame is transported to the metro or backhaul network.
Interior network elements known as core bridges (CBs) serve as transit nodes and contain provisioned static forwarding entries. Primary and backup tunnels are configured to optimize service availability and enable predictable failover performance. These CBs merely parse the outer three fields of the frame, which appear as any other traditional VLAN-tagged Ethernet frame. Based on the backbone DA and backbone VID, the CB forwards the frame to the correct port.
Figure 4 illustrates this process. Two physically separate customer sites are interconnected using PBT tunnels. Edge bridge 1 (EB1) and Edge bridge 2 (EB2) encapsulate the PBT MAC header and forward customer traffic through the operator network. Primary and backup tunnels are pre-provisioned by the operator’s network management system. When a customer frame ingresses EB1, the backbone MAC address of EB2, the MAC address of EB1, the backbone VID (4001), and a configured unique Instance Service Identifier (10,000) are added to the front of the frame. The PBT frame is forwarded to CB1. CB1 performs a simple lookup in the forwarding database and sends the frame to CB3. This process continues until the frame reaches EB2. EB2 removes the encapsulated MAC header and forwards the frame to the customer’s network.
Tunnels are monitored using the IEEE 802.1ag Connectivity Fault Management (CFM) control plane. Continuity check messages are sent periodically as a heartbeat message to ensure tunnel connectivity. Figure 4 shows a provisioned backup tunnel. CFM will detect an equipment or link failure, and automatically switch traffic from the primary tunnel to the backup tunnel.
PBT provides carriers with a variety of important benefits including predictable traffic engineering support, path load balancing, fast path protection switching, and path bandwidth management. Additionally, the MAC header encapsulation method is backward compatible with IEEE 802.1ad Provider Bridge devices. This lowers the overall capital expenditures required to deploy PBT-enabled service transport in existing and emerging operator networks.
Many emerging and planned operator networks exceed the traditional capabilities of Ethernet in terms of scalability, resiliency, and failover reliability and performance. PBT addresses these shortcomings by offering a native Ethernet technique for reliable and predictable service transport.
Despite the impressive advantages of PBT, an important limiting factor is the sophistication of the management system required to provision paths, tunnels, forwarding entries, and services. Other technologies, such as MPLS Virtual Private LAN Service (VPLS), use dynamic control protocols to reserve bandwidth, automatically choose optimal paths, and exchange tunnel and virtual circuit (service) information. While popular in many core networks, MPLS VPLS technology would be cost-prohibitive in many access and aggregation network applications due to the large capital expenditures resulting from the complexity and expense of MPLS router hardware and software.
PBT proponents believe that comprehensive element and network management systems are essential for large-scale, robust Ethernet service delivery. This investment can be leveraged and extended to provide PBT provisioning that enhances the operator’s control and utilization of the network.
Occasionally, new technologies that make an initial industry splash become relegated to niche applications. Is PBT destined for the same fate? PBT supporters believe a variety of indicators point to promising, mainstream applications. Early trials and deployments of these applications are already underway. Consider the following:
- PBT is a natural evolution of Ethernet. It employs MAC header encapsulation and reuses the well known, venerable Ethernet frame format.
- PBT adds compelling connection-oriented services to Ethernet.
- PBT can coexist with traditional RSTP/MSTP environments by logically splitting the VLAN identifier space.
- PBT complements MPLS core networks by extending connection-oriented transport into large-scale metro and access networks.
- PBT has emerged as an ideal technology for 3G/4G packet-based mobile backhaul applications, with promising field trials underway.
- Multiple vendors are responding to growing carrier and operator interest by developing a range of network elements and management tools. A wide array of implementations creates healthy competition and ensures attractive economics.
- PBT is being formally specified under the auspices of the esteemed IEEE 802.1 community of standards developers.
PBT is poised to be the next significant evolution of Ethernet technology. Carriers and operators that want to expand their access and metro networks cost-effectively are looking at PBT for greater scale, resiliency, and control.
PBT is becoming a more important and more relevant technology as the industry is quickly moving toward an international, interoperable standard; it is anticipated that PBT will become an IEEE 802.1 standard within approximately 24 months. As the standards coalesce and multiple vendors begin offering PBT-enabled network elements and management tools, carriers and operators are gaining confidence in trialing the technology for challenging network configurations. In 2007, several domestic and international initiatives will reinforce the belief that PBT will change the telecommunications landscape.
Kevin Daines is chief technology officer at World Wide Packets (www.wwp.com). His background in Ethernet technology includes more than 15 years of experience in engineering and networking. Since 1996, he has been an active member of the IEEE 802 Local Area Network and Metropolitan Area Network Standards Committee. He has since received two Outstanding Contributor Awards for his work on the Gigabit Ethernet and Ethernet in the First Mile (EFM) standards. A former editor of the IEEE 802.3ah EFM OAM, Daines currently chairs the Congestion Management and Frame Expansion Task Forces.