Building for the Future Tata Communications Ethernet Evolution

Size: px

Start display at page:

Download "Building for the Future Tata Communications Ethernet Evolution"

Jeremy Warren
5 years ago
Views:

1 Building for the Future Ethernet Evolution A PBB Technology Briefi ng September 2010

2 Table of Contents Contents Background The Right Choice Future Proofing the Network VLAN s MAC Addresses Megabytes A Focus on Services Introducting Provider Backbone Bridges Layers Lead to Efficiency The PBB Advantage Today s Problems, Tomorrow s Solutions Endnotes

3 Building for the Future Background Ethernet product management executives and the Ethernet engineering team gathered in July 2008 in Montreal, Canada to discuss technology. Fate may have played a part in this impromptu meeting as all the right people, landed in the right place, at the right time. The meeting lasted approximately 20 minutes; however the outcome will impact the company and its customers for the next decade. In the meeting, participants agreed that needed to make a bold move in regard to its Ethernet technology portfolio. Different options were debated, according to time-tomarket considerations and financial risk. The easy path would be to follow the market and embrace a current technology with well-known capabilities. Ethernet engineering team argued that new technologies are available that could potentially transform the Ethernet services landscape. Meeting attendees agreed that true industry leaders must develop a broad view, complete with associated risk, to consistently deliver more than the market expects. This decision prompted a two-year search with an established goal of identifying a game-changing technology that would deliver cutting edge Ethernet services for the next seven to ten years. Today, introduces the first phase of a technology upgrade that promises to revolutionize the global and Indian Ethernet markets. The Right Choice Ethernet engineering team evaluated all present and future technologies both on paper and in the Ethernet laboratory in search of the best possible platform. Discussions and debates led to a single conclusion: Provider Backbone Bridging (PBB) provided the best-fit technology for both regional and global networks. This decision was based on choosing a platform that would help future-proof the network while expanding and enhancing service offerings. Future Proofing the Network Current Ethernet business, plus the tremendous growth predicted by industry experts, led Tata Communications to choose PBB technology because its innate design handles increases in VLANs and MAC addresses. 3

4 Building for the Future VLANs Current technology is restricted by standards to 4094 VLANs (or services). This concern grows more pressing with the fact that many customers require more than a single VLAN. PBB is designed to handle 16.7 million services a significant order of magnitude over the traditional technology using VLANs. MAC Addresses Industry experts predict 40 percent of all Ethernet sold will be multipoint services by This fact, coupled with the tremendous predicted growth of Ethernet, leads experts to predict that the number of customers eventually using multipoint services could be staggering. Multipoint services utilize MAC addresses and MAC address lookup tables to move frames around the network. A service provider must use a technology to efficiently manage MAC addresses and MAC lookup tables or their services will be impacted if the number of MAC addresses they manage on the network grows exponentially. PBB eliminates the exponential growth problem, greatly improves the efficiency of MAC addresses and lookup tables plus delivers greater security for the end customer. Megabytes The tremendous growth of the Ethernet market along with the ever increasing demand for bandwidth will force service providers to accommodate large bandwidths. is prepared on two levels. Through cable ownership, can tap large quantities of bandwidth on a wholly owned network within India and wholly owned undersea cables that literally encircle the globe. Secondly, chose equipment that handles 240 Gig today with the ability to handle 480 Gig within two years. A Focus on Services While technology upgrades capture attention, new technology services create business opportunity and customer value. Services are a significant focus for and important service announcements will be released over the next few years. technology evaluation focused on the ability of the technology to: Support and enhance existing services without changing these services Offer cutting edge services for years to come, and Create new services that meet a high standard of quality Introducing a new technology is never easy. The focus of this technical briefing is to provide market information on switch to PBB. The following excerpt from a Cisco Systems document clearly explains Provider Backbone Bridging technology. Introducing Provider Backbone Bridges Extracted from Provider Backbone Bridging and MPLS: Complementary Technologies for Next-generation Carrier Ethernet Transport, authored by: Samer Salam and Ali Sajassi, Cisco Systems With roots as a Local Area Network (LAN) technology, Ethernet initially lacked the flexibility, security, and scalability of carrier technologies. In the beginning, Ethernet offered a shared transport medium where all end-devices were connected and shared a common broadcast domain. Later on, IEEE standard 802.1Q defined the concept of a Virtual LAN (VLAN) as a means of creating independent logical LANs, with disjoint broadcast domains, over a single physical network. The 802.1Q standard extended the Ethernet frame format to include a 4

5 Ethernet Evolution VLAN tag (Figure 1 frame format b), which was added by switches, to identify the frame s Virtual LAN over inter-switch links. A switch would admit and forward frames only on ports that were configured with the same VLAN identifier. The motivations for VLANs were two-fold: providing security over a shared physical network through logical traffic segregation and more efficient bandwidth utilization by limiting the scope of flooded broadcast or multicast traffic to a VLAN, rather than the entire physical network. This was developed in the context of modest sized LANs where a few hundred VLANs were sufficient to address most, if not all, deployments. Given those constraints, IEEE 802.1Q converged on assigning 12 bits for the VLAN identifier. This limited the number of identifiers to The standard reserved identifiers 0 and As service providers began exploring Metro Ethernet services, they leveraged 802.1Q to offer services to end customers over bridged networks. A service provider assigned each customer service instance its own VLAN identifier to guarantee traffic security and optimal bandwidth utilization. This limited the maximum number of service instances that a provider could offer over a network to 4094 the number of allowed VLANs. This scalability limitation prevented service providers from adding services and increasing revenue on installed networks. In addition, 802.1Q did not provide a mechanism to distinguish the customer VLAN identifier space from the service provider identifier space. For example, if a service provider offered an inter-city Transparent LAN Service to an enterprise, and that enterprise required some number N of VLANs then the service provider had to assign these enterprise N VLAN identifiers for that single service. This implied that the service provider had to coordinate the identifier space amongst its customers in order to prevent two customers from using the same VLAN a matter that introduced operational complexity. IEEE standard 802.1ad (which is now part of IEEE standard 802.1Q-2005) solves this problem by introducing a Service VLAN tag which is pre-pended to the 802.1Q VLAN hereafter referred to as the Customer VLAN in the Ethernet frame (Figure 1, frame format c). The Service VLAN tags are added by provider edge bridges to customer frames as they ingress into the service provider s network. The tags are removed from those frames, again by provider edge bridges, as the frames egress the provider s network. The frames are bridged within the provider s network based solely on the Service VLAN. As such, the Customer VLANs remain invisible to the service provider. 5

Building for the Future 2008 IEEE Communications Magazine Figure 1 - Various Ethernet frame options The Service VLAN serves the purpose of uniquely identifying the customer service instance in the

6 Building for the Future 2008 IEEE Communications Magazine Figure 1 - Various Ethernet frame options The Service VLAN serves the purpose of uniquely identifying the customer service instance in the service provider s network, and allows for the transparency of Customer VLANs within the carrier s network. For example, in a network employing 802.1ad Ethernet technology, a service provider may use Service VLAN 100 for a Transparent LAN Service offered to Enterprise Y, and Service VLAN 200 for another service to Enterprise Z. Enterprise Y has five Customer VLANs (11 through 15) whereas Enterprise Z has 10 Customer VLANs (11 through 20). Note how the two customers have overlapping Customer VLANs (11 through 15). This would not have been possible without 802.1ad technology. It is worth noting that IEEE 802.1ad does not solve the problem of service instance scalability: The Service VLAN field remains 12 bits in width. Thus, service providers are still limited to a maximum of 4094 service instances per bridged network. Furthermore, with IEEE 802.1ad and the original IEEE 802.1Q, service provider equipment operating at the Ethernet service layer has to learn the MAC addresses of customer devices, a matter that poses yet another scalability challenge. Because the service provider s equipment is bridging customer frames, they have to learn the addresses of either the customer edge devices (when those devices 6

7 Ethernet Evolution are routers) or the addresses of the customer end-stations (when the customer edge devices are bridges). Disabling address learning is not an option, especially in the context of multipoint services, because it leads to excessive flooding of unknown unicast frames. This, in turn, wastes network bandwidth. Service providers are thus bound by the size of their equipment hardware address tables when attempting to scale the number of customers or customers sites over their existing networks. For example, consider a service provider offering a Transparent LAN Service for an enterprise with 50 sites and 200 end-stations per site. For this service, the service provider s equipment has to learn 10,000 addresses (50 x 200). If the provider were to accommodate only 10 such customers, the provider s devices would each have to learn 100,000 address entries. This exceeds the capability of most Ethernet switches on the market, as typical state-of-the-art Ethernet switches have MAC address table sizes ranging between 4,000 and 64,000 entries. IEEE draft standard 802.1ah, or Provider Backbone Bridges, aims to solve the service instance and MAC address scalability problems highlighted above. PBB extends IEEE 802.1ad and introduces a hierarchical network architecture that allows service providers to build large bridged networks. In this model, networks of IEEE 802.1Q bridges are aggregated into networks of IEEE 802.1ad Provider Bridges which, in turn, are aggregated into a network of Provider Backbone Bridges. Figure 2 depicts this hierarchy in an example network IEEE Communications Magazine Figure 2 - Provider backbone bridging network heirarchy 7

8 Building for the Future To solve the MAC address scalability problem, PBB introduces a new frame format (Figure 1, frame format d) that provides a MAC tunneling encapsulation scheme customer Ethernet frames are encapsulated within provider Ethernet frames as they ingress the PBB network, thereby hiding the customer addresses from the PBB core. Devices in the core of a PBB network forward traffic based on backbone MAC (B-MAC) addresses. This effectively confines the need to learn customer addresses to the edge devices of the PBB network these edge devices are called backbone edge bridges. A given backbone edge bridge is required to learn the addresses of only those customers that it supports, and a given core device is required to learn the addresses of only the backbone edge bridges (as opposed to having to learn addresses of all end customer devices). This greatly enhances solution scalability. For example, in Figure 2, assume that a service provider employing 802.1ad access and PBB core networks is offering a Transparent LAN Service to a customer with two sites corresponding to CE1 and CE2. Furthermore, assume that 200 customer end-stations (H1 through H200) are sitting behind customer edge bridge CE1, and another 200 end-stations (H201 through H400) are behind customer edge bridge CE2. The provider devices transporting service frames between the two customer sites, which employ 802.1ad technology (for example devices PEB1, PB1 and PEB2), will end up learning 400 customers addresses corresponding to all end-stations. Now shifting focus to the PBB section of the network: backbone edge bridge BEB1 will learn customer addresses for the directly connected site, i.e. the addresses corresponding to hosts H1 through H200, in addition to the customer addresses for the remote site of the service, corresponding to hosts H201 through H400 (a total of 400 addresses). Similarly, backbone edge bridge BEB2 will learn a total of 400 addresses. However, the bridges in the core of the PBB network will only have 2 MAC address entries in their forwarding tables for that service: these correspond to the backbone addresses of devices BEB1 and BEB2. This is two orders of magnitude less than the MAC table size of the 802.1ad bridges. The occurs because BEB1 and BEB2 encapsulate the customer frames with PBB, using their own backbone addresses in the source address field, before forwarding those frames to the PBB core. To solve the customer service instance scalability problem, the PBB frame format introduces a 24-bit service identifier field known as the I-SID. Each customer service instance is assigned a unique I-SID value, which is global within a network administrative domain. Hence, the new PBB frame format effectively expands the number of service instances from 4094 to a theoretical maximum limit of roughly 16 million (224). Note that I-SIDs are visible to backbone edge bridges only, and are transparent to the devices in PBB network core. The PBB frame format also includes a 12-bit backbone VLAN identifier (B-VLAN). This allows the service provider to partition their PBB network into different broadcast domains for efficient bandwidth utilization, especially in multicast applications. The B-VLAN also enables the service provider to perform load-sharing within the PBB network core by means of bundling different I-SIDs into distinct B-VLANs and mapping the latter into different spanningtree instances. PBB can be seen as a natural evolution of 802.1ad. For a service provider to upgrade existing 802.1ad networks to PBB, they only need to upgrade the edge devices to implement backbone edge bridge functionality. The core of the 802.1ad network remains unchanged. Internally, in the broadest sense, a backbone edge bridge comprises two types of building blocks: one or more I-components and a single B-component. The I-component connects to the customer 8

Ethernet Evolution equipment and is responsible for learning customer addresses, encapsulating customer frames with backbone addresses, and adding the I-SID.

9 Ethernet Evolution equipment and is responsible for learning customer addresses, encapsulating customer frames with backbone addresses, and adding the I-SID. The B-component connects to the PBB core devices and adds the B-VLAN and bridging backbone-encapsulated frames based on backbone addresses. Figure 3 shows the model IEEE Communications Magazine Figure 3 - IEEE 802.1ah backbone edge bridge model Layers Lead to Efficiency will build a layered Ethernet network, with each layer specific to a geographic area and/or a function. The goal is to create an efficient network that can deliver a wide variety of services. This network will consist of the following four layers: 1. Core: A PBB network that encircles the globe, creating an Ethernet highway connecting all major regions. The purpose of this layer is to move frames efficiently around the globe. 2. Regional: A PBB network within a region designed to offer services within a region and to move international traffic to and from the Core. The focus of this layer is on services. 3. Metro: A network within a city that offers services within the metro and provides connectivity from customer premises to regional networks. Equipment and technologies are currently under evaluation. 4. Access: Point of presence (PoP) outlier location that is not part of any particular network layer. The purpose of Access is to move traffic from the outlier PoP to the Metro, Regional, or Core layers. 9

10 Building for the Future The PBB Advantage will offer services utilizing the PBB technology that greatly benefit customers and promise to uphold the company s commitment of quality and variety to our customers. The following summary points explain plans for introducing new and existing services or options that coincide with the introduction of PBB: All existing services will continue to be delivered on the new technology existing services will not change. Protected ports will become an option so customers can elect to prevent an outage due to a port or port board failure. will enhance all SLAs. bandwidth offerings will increase 1,000 Meg to 10,000 Meg in 500 Meg increments using 10 GigE ports. Security will improve since both the customer VLAN and the customer MAC address will be hidden under the PBB wrapper. Additional services and options will be announced based upon PBB s abundant capabilities. Today s Problems, Tomorrow s Solutions Leadership requires innovative companies and their employees to do more, reach farther, and think faster. Although investment in new Ethernet technology is the right choice, the path to innovation is sure to pose difficulties. Hard work lies ahead and the first order of business is to communicate the benefits of PBB technology to current and potential customers. The purpose of this technology brief is to explain Tata Communications rationale in adopting this new core technology. With this understanding, both present and future customers will undoubtedly realize that made a sound technology investment at the right time. is making a quantum leap to PBB a technology that is already recognized as important to the future of telecommunications. The industry is still debating how PBB will work with existing technologies, causing delays for other service providers to deploy PBB. PBB deployment does not involve integration with any other technology. Eliminating integration issues and moving to a single technology will simplify operations and create a high quality, global network. Excitement is growing as prepares to offer new services and features with a new network that promises to challenge the marketplace. As a proven leader in delivering Ethernet services, is making a bold, calculated move that will positively impact the future of Ethernet and ultimately benefit customers. 10

11 Ethernet Evolution Endnotes 1 V. Rajaravivarma, Virtual local area network technology and applications, in Proceedings of the Twentyninth Southeastern Symposium on System Theory, March 1997, pp S. Halabi, Metro Ethernet, Cisco Press, October P. Wang, C. Chan, and P. Lin, MAC Address Translation for Enabling Scalable Virtual Private LAN Services, Proc. 21st Int l. Conf. Adv. Info. Net. and Apps. Wksp., vol. 1, May 2007, pp A. Elangovan, Efficient Multicasting and Broadcasting in Layer 2 Provider Backbone Networks, IEEE Communications Magazine, vol. 43, no. 11, November 2005, pp IEEE P802.1ah/D3.6, Virtual Bridged Local Area Networks Amendment 6: Provider Backbone Bridges, June

12 India Limited C-21/C-36, G Block, Bandra Kurla Complex, Vidyanagari Post Office, Mumbai , India Tel Fax Asia 5 Shenton Way #34 10 UIC Building Singapore Tel Fax Bank of America Tower 12 Harcourt Road Central, Hong Kong Tel Fax Australia King Street Wharf Suite 503, 35 Lime Street Sydney NSW 2000 Australia Tel Fax North America 2355 Dulles Corner Blvd 7th floor Herndon, VA 20171, USA Tel Fax Matawan Rd 3rd Floor Matawan, NJ 07747, USA Tel Fax Rue Carrie-Derick Montréal (Québec) Canada, H3C 6W2 Tel Fax Middle East & North Africa Hamdan Street, City Center Building Block A, 2nd Floor, Office # 204 P.O. Box Abu Dhabi, United Arab Emirates Tel Fax Europe Exchange Tower, Suite Harbour Exchange Square London, E14 9GE, England Tel Fax Avenue Charles de Gaulle Neuilly sur Seine France Tel Fax Avenida de Europa 4, Bajo A Parque Empresarial La Moraleja Alcobendas Madrid, Spain Tel Fax Bettinastrasse 30 D Frankfurt am Main Germany Tel Fax , a member of the $62.5 billion Tata Group, is a leading global provider of a new world of communications. The emerging markets communications leader leverages its advanced solutions capabilities and domain expertise across its global and pan-india network to deliver managed solutions to multi-national and Indian enterprises, service providers and Indian consumers. range of services include transmission, IP, converged voice, mobility, managed network connectivity, hosting and storage, managed security, managed collaboration and business transformation for global enterprises and service providers, as well as Internet, retail broadband and content services for Indian consumers. The Tata Global Network encompasses one of the most advanced and largest submarine cable networks, a Tier-1 IP network, with connectivity to more than 200 countries across 300 Pops, and more than 1 million square feet of data center and co-location facilities. unique emerging market depth and breadth of reach includes a national fiber backbone network and access to network in over 60 cities and 125 Pops in India, strategic investments in South African converged services operator, Neotel, Sri Lanka and Nepal and, subject to fulfillment of conditions precedent, a 50% ownership in China Enterprise Communications (CEC) providing full country VPN coverage in China. Servicing customers from its offices in over 80 cities in 40 countries, Tata Communications is the number one global international wholesale voice operator and number one provider of international long distance, enterprise data and Internet services in India, the Company was named Best Wholesale Carrier at the World Communications Awards in 2006, Best Pan-Asian Wholesale Provider at the 2006 and 2007 Capacity Magazine Global Wholesale Telecommunications Awards and was awarded Best Progress in Emerging Markets at the 2008 Mobile Communication Awards. Limited is listed on the Bombay Stock Exchange and the National Stock Exchange of India and its ADRs are listed on the New York Stock Exchange. (NYSE: TCL) Ltd. All Rights Reserved. 4p-Fiber-Channel-Connectivity-v1.0-august-2010

National Dedicated Ethernet Service

National Dedicated Ethernet Service Flexible, reliable service to over 120 PoPs in India, and access to the global network www.tatacommunications.com 1 Connecting througout India Or, Reaching to the world