MPLS: Paving the path to Unified Communications

White paper
August 2009

By Greg Banks

Executive summary

An emerging class of business application is breaking down the traditional barriers between telephony, video and data services. These Unified Communications (UC) applications are increasing productivity, driving down costs and augmenting the quality and ease of communication in organizations from sea to sea. What these applications share is a need for any-to-any connectivity and guaranteed quality of service (QoS).

Eliminating latency in data delivery is key to assured QoS and one of several significant reasons why the majority of Canadian organizations have made the switch from layer 2 data transfer protocols such as frame relay, Ethernet and asynchronous transfer mode (ATM) to multi-protocol label switching (MPLS)-enabled IP virtual private networking (IP VPN). MPLS is an IP data transfer protocol developed more than a decade ago. By expediting data packet forwarding via label switching rather than by address lookup and forwarding, it greatly reduces latency in delivery.

MPLS is not new technology, but because its latency rate is so low, it is the subject of renewed interest as the obvious protocol for use with newer, high-bandwidth applications. MPLS has largely replaced layer 2 transfer protocols in carriers' core networks and within Canadian enterprise-level organizations.

As a protocol that enables IP VPN, MPLS can be offered as a managed solution that reduces both the need for infrastructure and the burden placed on the IT team. In this way, MPLS lowers the total cost of ownership (TCO) of organizations' networks. MPLS also lowers operating expense in other ways: through economies realized by moving to one network for voice and data; by lowering service and telephony costs through shared infrastructure and any-to-any communication; by greatly increasing ease of scalability; and by enabling productivity gains, reduced latency and other efficiencies realized through the use of convergence applications.

Perhaps the greatest overall benefit that MPLS brings to organizations is the enabling of new convergence applications that boost productivity and reduce latency in business. This paper explores the top five such applications in detail and looks to the future of application use in the organization. Its purpose is both to elaborate the many benefits of MPLS in the organization and to set out best practices in making the switch from layer 2 technologies.

MPLS explained

MPLS is an IP data transfer protocol developed more than a decade ago. It was designed to expedite packet forwarding via label switching, rather than by address lookup and forwarding. MPLS integrates a label swapping and forwarding mechanism with IP routing, improving the speed and scalability of the IP core. It is the ideal data transfer protocol for use with the high bandwidth, QoS-enabled applications that have begun to proliferate in the business world. Since the protocol's significant traffic engineering capabilities are far superior in handling data flows such as voice and video, it has largely replaced layer 2 transfer protocols such as frame relay and asynchronous transfer mode (ATM) in carriers' core networks.

In a traditional layer 3 network, routing is based on IP addresses such as 192.68.121.102. Every router encountered along the IP data packet's journey strips the layer 2 address from the packet, examines its layer 3 IP address and determines the next hop based on its routing table. Once the next hop is determined, the IP packet is given a new layer 2 address and the packet is forwarded to the next router. This hop-by-hop approach to data routing introduces latency and eats up processing time. This means that costly upgrades are needed to accommodate large, complex networks.

Since an MPLS-enabled network substantially reduces packet-processing requirements, latencysensitive applications such as voice and video can be delivered in near real time. The way that this is accomplished is by replacing the layer 3 look-up function of a traditional network with a label-swapping mechanism, enabling low-latency routing.

Since an MPLS-enabled network substantially reduces packet-processing requirements, latencysensitive applications such as voice and video can be delivered in near real time. The way that this is accomplished is by replacing the layer 3 look-up function of a traditional network with a label-swapping mechanism, enabling low-latency routing.

Instead of building a road for each car, MPLS affixes an information-rich label to it, influencing its behaviour as needed for QoS requirements – express lane or collectors, where to exit and so on – and sends it out on the highway. This any-to-any connectivity is coupled with intelligent traffic engineering. In this way cars with blue labels know to enter the highway at Montréal and stay in the passing lane until they exit at Ottawa, whereas red-labelled cars that entered the highway at Québec continue on to Toronto in the middle lane. Labels are quickly checked at points and updated as needed thanks to automated traffic engineering, for example to assign alternate routes in the case of congestion.

Technically speaking, MPLS labelling reduces latency by providing an implicit destination across the core, much like a permanent virtual circuit (PVC). At each router the label is examined and swapped with the next destination label. The IP address is hidden from the core network as the packet traverses it.

MPLS addresses routing limitations because:

• MPLS-enabled routers make forwarding decisions very quickly on the basis of the label, rather than consulting a layer 3 routing table to determine the next hop
• A complete path is determined in advance for a given traffic flow, and packets in the same flow always follow the same path, minimizing transit delays

The following chart illustrates a common layer 2 interworking functionality (Ethernet) as compared to MPLS-enabled IP VPN.

Chart illustrates Ethernet Interworking functionality as compared to MPLS-enabled IP VPN

Benefits and features of MPLS

The decision to switch to MPLS is usually driven by a combination of two or more of the following drivers:

• Deployment of convergence applications
• Network expansion
• Need to maximize the capabilities and reliability of applications
• Contain or reduce costs

Some of the key business benefits that MPLS brings to the network are scalability and consolidation of multiple systems for voice and data. Both of these result in a lower total cost of ownership (TCO). But without question the greatest technical benefit brought to the network is the ability to support other applications beyond IP such as QoS and native multicasting. Legacy networks simply do not have the reach, any-to-any connectivity and full QoS that would allow organizations to fully leverage today's applications.

Key business benefits of MPLS

• Key business benefits of MPLS
• Lower operating expense by moving to one network for voice and data
• Lower equipment, service and telephony costs with shared infrastructure and any-to-any communication
• Enable productivity gains and efficiencies realized with convergence applications
• Ease of scalability: add nodes to the network without reconfiguring existing nodes

Top features of MPLS

Scalability and ease of connectivity

Because MPLS supports IP (layer 3) routing, data reaches its destination as it does on the Internet – without the need for a predefined physical or virtual path. With traditional WAN layer 2 protocols, a path must be manually configured between endpoints – an onerous task in a full mesh network and one that doesn't lend itself to scalability. With MPLS, each information packet finds its own way to the intended recipient via the most efficient means possible. Since this is accomplished through label switching, with the IP address hidden from the core network, there is no need to build a logical connection or to change the corporate IP addressing structure for adoption.

Organizations employ any of a number of local loop transport methodologies, based on bandwidth and anticipated service level requirements. Optical Ethernet (OE) is generally preferred for bandwidth requirements exceeding 5 Megabytes per second (Mbs), whereas T1 and bonded T-1 are normally used for requirements of 1.5 Mbs to 3 Mbs. In the case of remote branches with reduced bandwidth requirements, connectivity may be provided via a dedicated Digital Subscriber Line (DSL) or over the Internet using Internet Protocol Security (IPsec) over DSL.

Simplified configuration

The any-to-any nature of MPLS saves both on the cost of initial set-up and of ongoing configuration adjustments. Unlike layer 2 protocols where a point-to-point communication protocol had to be set up individually between each node and every other node with which it needed to communicate, MPLS connects all authorized nodes, allowing at-will network expansion.

Custom topology

MPLS lends itself to adaptation. It is a uniquely flexible protocol that can accommodate a variety of needs. For example, network connection speed is highly scalable – from sub T-1 (less than 1Mbs) up to gigabit Ethernet speeds.

It is possible also to configure separate virtual private networks (VPNs), by line of business or extranets for external partner connectivity. If, for example, you wanted to isolate the 50 marketing department locations from the finance department's, separate VPNs can easily be set up using unique virtual routing forwarding identifiers (VRF).

Ensure end-to-end quality of service

As noted earlier, one of the benefits of MPLS is the ability to support a variety of other IP applications. To ensure QoS, network administrators to prioritize and cue traffic as desired, using multiple class of service (CoS) queues. Priorities are recognized by the network core as well as by the edges. In this way, voice and video data streams can be prioritized over less time-sensitive data such as e-mail, resulting in a better service level (i.e. less delay, jitter and packet loss) for priority traffic. This provides a near-real time video and toll quality audio experience.

Increased availability

MPLS had increased the robustness of communication – so much so that 71 percent of North American firms with over 500 employees have adopted it in their wide area network (WAN). High availability is achieved through a number of mechanisms. On the edge, customers can subscribe to different levels of access. Dual customer edge (CE) devices and provider edge (PE) router diversity are combined with advanced routing protocols such as hot standby routing protocol (HRSP) and border gateway protocol (BGP) to ensure that a failure on one edge, either CE or PE, does not impact data connectivity at that location. On the provider side, dual homing of PE routers to a fully meshed core provider router (P router), for example, greatly enhances survivability.

Efficient bandwidth utilization

Combining end-to-end QoS with a conversion of legacy applications to IP will help to save on bandwidth requirements. Application acceleration products are also gaining in popularity as a means of analyzing and prioritizing traffic. Such products make it possible to shape traffic based on application type and compress for acceleration across the network, rather than increasing bandwidth to boost application performance.

Save on resources

In the case where a provider offers a fully managed service, such as with Bell's IP VPN, the CE is owned and managed by the provider, who bears the burden of WAN network management. This allows scarce IT resources to concentrate on business-critical applications and support. Additionally, there is no capital outlay required for edge equipment, nor for ongoing evergreening, maintenance or upgrade expenses.

MPLS: Paving the path to Unified Communications

Whether they reduce latency or increase productivity by other means, the ability to leverage the applications that ride on top of MPLS is often reason enough to make the switch. Following are the five most important MPLS-enabled convergence applications currently making waves in the business world:

PSTN SIP trunking

A Session Initiation Protocol (SIP) trunk connection enables voice over Internet protocol (VoIP) telephone calls to any national destination through your installed private branch exchange (PBX). SIP trunks share the same local access connection used for your data services. SIP trunking also eliminates the need for local public switched telephone network (PSTN) gateways and interfaces such as Basic Rate Interfaces (BRIs) and primary rate interfaces (PRIs).

Why SIP trunks? In addition to cost savings, SIP trunks will become an important enabler of advanced UC features such as work-at-home over an existing residential line, configurable incoming call routing rules, multi-user white boarding and mobility integration. SIP trunking will also deliver a better audiovisual experience. Digital encoding combined with best-in-class Quality of Service (QoS) in an MPLS core has no trouble delivering excellent sound quality and rich information such as caller images, collaboration sessions and presence details.

Native IP multicasting

Native multicasting is an enabling technology. It is a one-to-many data transfer protocol that promotes more efficient use of the WAN for a greater number of applications. It does this by reducing bandwidth and server requirements while obviating the need for significant manual configuration. Multicasts require data to be sent only once, even if it is to be delivered to a large number of recipients. The nodes immediately upstream of recipients replicate the packet as needed. In this way, multicasting scales to a larger receiver population by not requiring prior knowledge of who or how many receivers there are.

Native multicasting, as opposed to generic routing encapsulation multicasting, does not require intensive manual configuration between sites. This greatly reduces setup costs while taking full advantage of the any-to-any functionality of MPLS. The single greatest use of multicasting today is webcasting: with multicasting, expensive generic routing encapsulation (GRE) tunnel setup is no longer required. One of the most promising applications is employee training: getting rich media materials out to a large number of recipients without the expenses of printing, the cost of sending people to a training location, or involving a tremendous quantity of bandwidth.

IP videoconferencing

The muddy interface and complicated controls of yesteryear are gone, replaced with the crystal clear image and sound of digital videoconferencing in high definition (HD). Thanks to Internet protocol and a QoS-enabled MPLS core, videoconferencing is far less of a bandwidth hog and the sound and picture quality are substantially enhanced. Videoconferencing comes in conference room, high-end telepresence and desktop varieties. It is one of the cornerstones of Unified Communications, now in use as an everyday communication medium.

Although many firms use reduction in travel as a key business driver for IP videoconferencing, the real value lies in bringing key resources together in several different meetings on the same day. High quality videoconferencing comes in conference room, high-end telepresence and desktop varieties. It is one of the cornerstones of Unified Communications and is now being used as an everyday communication medium.

Site-to-site WAN VoIP

Wide area networks (WANs) can span a city or the entire world. What they all share in common is a need for continuous, uninterrupted connectivity and efficient data transfer. MPLS provides a fully meshed survivable core for WAN VoIP networks while ensuring that QoS is strictly maintained. The result is richer applications running across the network with less latency and more predictable performance, at a reduced cost.

IP/PBX desktop VoIP

Desktop VoIP enables connection to any extension on the corporate network without incurring long distance fees. It comes with a host of other advanced features such as automated attendant, find me - follow me functionality and conference calling. It also reduces latency and boosts productivity by transparently routing office calls to a home or mobile telephone, sending voice mail to email or alerting callers of 'busy' status before a call is placed. Other applications include allowing customer service representatives to see customer contact and account information when a call is received, contact database-to-call functionality and more.

Once a VoIP PBX service is in place, moving an employee phone is a simple matter of unplugging and switching. The phone number moves with the user, minimizing configuration requirements. With Web and video conferencing, users can see the person that they are talking with for a more personalized experience.

Other functionality scenarios include allowing customer service representatives to see customer contact and account information as soon as a call is received, one-touch calling from a contact database and communication media upgrades from Internet messaging to voice to video.

Case study: MPLS success

The past decade has seen many successful MPLS implementations in Canada. Following is an example that serves to illustrate the advantages that the technology brings to specific applications.

Case study: CIBC's combined voice and data switch

Challenge

Between 2002 and 2008, the Canadian Imperial Bank of Commerce (CIBC) turned a challenge into an opportunity. The bank found itself challenged both to efficiently meet current communications needs and to grow in coming years.

Key issues included:

• Managing multiple data network cores
• Inefficient use of network resources
• Duplication of tasks, extensive manual processes
• Limited communications functionality at the desktop
• Non-standard processes, especially at the end-user level

Instead of seeing a multitude of problems, CIBC board-level decision makers saw an opportunity for a major network refresh. So instead of the intuitive quick fix involved in standardizing voice traffic systems and moving to VoIP, they decided to tackle data and VoIP together, increasing operational efficiencies and laying the foundation for advanced IP-based communications for the next decade and beyond.

Solution

Five proposals were tendered, each including two technology vendor options and two management solution options for a total of 10 potential deployment scenarios. Bell was ultimately chosen to replace multiple legacy frame relay and ATM data networks with an MPLS-enabled IP VPN core. There were three parts to the solution:

• Combining multiple networks into one
• Deploying an upgraded access network within the campuses and branches
• Installing a real-time communications (RTC) infrastructure within data centres, including VoIP telephony

Bell was also to deliver 99.9999 percent uptime, or less than one minute per year of downtime – down from just under an hour. Of CIBC's 20,000 connections in need of an upgrade, half were be SIPenabled, with the following functionality:

• Presence features
• Audio conferencing
• Equal split between traditional hard phone and soft phone
• Enhanced voicemail with biometric voice recognition security

Results

Overall results included greater operational cost savings, scalability, security management, adaptability, and risk reduction. Specific infrastructure and network management benefits included:

• Improved inventory management of network assets
• Consolidation and convergence of network infrastructure, positively impacting staffing, travel, and management costs as well as providing a consolidated view of network assets
• 15 to 25 percent reduction in telephone system administration thanks to IP phone plug-and-play
• Savings on long distance toll charges
• Improved security of the network environment
• Deployment of QoS tools to allow for greater control over network prioritization and performance
• Increased IT staff productivity
• Consistency of user interface and experience with choice of hard (desk) phone and soft phone
• Ability to collaborate cross-functionally and across disparate locations

The future of data transfer: application acceleration

Where is MPLS going next? Will something replace it? When considering a major infrastructure investment, it's tempting to think that holding off a bit longer will yield an even better solution. After all, technology solutions evolve so quickly. The key thing to remember with MPLS is that it is simply a data transfer protocol – one to which over 70 percent of Canadian organizations have migrated their networks. While the protocol is not likely to change, what will continue to evolve are the applications and functionality that travel over MPLS.

This paper has described several applications that are now being adopted by organizations thanks to MPLS. The next most exciting development surrounding the technology is a new way of optimizing the networking of these and other applications by focusing on the applications rather than on the network itself. It's called Application Acceleration Networking (AAN).

AAN addresses a common issue: when an application's response time slows down due to high bandwidth use, a typical reaction is to increase bandwidth. The problem with such a response is that bandwidth-intensive applications will simply grab even more bandwidth, so the need continues to grow. Deploying AAN curtails rapid bandwidth growth by intelligently prioritizing it for any number of application buckets. AAN is essentially QoS for applications. In some implementations, the AAN software is intelligent enough to be aware of all application flows and packet priorities across a network in real time. When this is the case, lower priority traffic is only sent from A to B when there is sufficient bandwidth available, even if those two points are not themselves engaged in sending higher priority traffic.

As an example, suppose that site A and site B have multiple voice flows between them, consuming 90 percent of available bandwidth. Site C wants to send a large email to Site A. However, as site C is aware that the higher-level voice traffic is consuming almost all available bandwidth, it holds off sending or reduces its transmission rate until sufficient bandwidth is available.

A custom dashboard and reporting functionality are typically provided to enable the IT team to monitor network and application performance. The team can then fine-tune the process, including increasing bandwidth when absolutely required.

Making the switch: Best practices in migrating to MPLS

Making the switch to MPLS is a major decision – one worthy of significant planning and due diligence. The first step is to outline what the overall deliverables should be. Think why you are making the move: is it primarily for reasons of scalability? Better application performance? Prioritizing the factors driving the switch should allow you to prioritize deliverables as well.

Common goals include:

• Making sure that there is no business interruption(need for redundancy)
• Getting the best possible performance of applications across the WAN (bandwidth considerations)
• Reducing infrastructure and ongoing costs (MPLS versus ATM, etc.)
• Selecting a vendor that employs best practices and is able to provide ongoing support and service

Implementation scenario considerations

Once you have committed to making the switch to MPLS, you will need to nail down and quantify service features, portfolio quality, access options and a host of other metrics. What will your MPLS network look like in terms of scale, reliability, flexibility and security? Following is a list of considerations you will want to make before settling on an implementation scenario:

• What impact might MPLS have on the network management system, especially surrounding accounting and billing? Tools such as NetFlow and IP accounting will cease to function with MPLS
• QoS requirements: What applications do you want to map to what quality of service? If for example IP phones are used, that traffic will need to be treated with the highest priority in the carrier network
• Diversity requirements: You may want to ensure redundancy through a number of links. That way, if one link goes down (for example there is a cable cut) you still have a backup. Carriers can typically provide half a dozen or more redundancy options, so be sure to choose the one that suits you best
• Bandwidth requirements: what bandwidth do you require per site?
• What are the memory requirements and are there any other potential upgrade requirements for devices in your network?
• Specify private network needs: If you need to keep the finance, marketing or other department traffic separate, you may need to specify a number of VPNs

Prior to implementation

• Inventory your applications and note any that require special treatment (i.e., non IP-enabled such as systems network architecture)
• Determine required software and firmware versions for all network devices, based on application requirements
• Before installing the target configuration, test it in a controlled environment to ensure that it will be stable
• Determine routing requirements (i.e. static or BGP)
• Work with suppliers' project management and engineering teams to establish an acceptance plan and cutover schedule

Implementation

• Upgrade network devices as required and ensure that the network is stable and has an established baseline for performance
• Run acceptance tests and verify against baseline
• Have a backout plan in place and ensure that all involved in the cutover have a copy
• Allow the network to stabilize
• Disconnect the old network

Working closely with your MPLS provider is important in ensuring a seamless implementation. They will typically be able to draw on significant experience and lighten the burden, not to mention greatly increasing the likelihood of a smooth transition. Choosing the provider that makes the most sense for you is perhaps the most important step involved in switching to a new network.

What to look for in a provider

Choosing the right provider can eliminate headaches down the road. Offerings can be very different, so it is important to evaluate each very closely. Here are some things to watch for:

• Core redundancy: how is the provider's network architected to ensure the availability meets your needs?
• What type of service level guarantees are available
  • Do they meet your requirements for availability
• Audit status: The gold standard in service organization auditing is the Statement on Auditing Standards No. 70 (SAS 70). At 6 months in length on a yearly basis, it is the most stringent audit in the marketplace and ensures that providers are adhere to a verifiable standard
• System hardening: are systems automated end-to-end, or is there significant room for human error? Hardening should encompass:
  • Service activation
  • Service monitoring
  • Reporting
  • Assurance
• Checks and balances: what kinds of checks and balances has the provider put into place to ensure that configurations are correct before you implement them?
• Is end-to-end system management offered? Outsourcing the management of your MPLS network makes sense for many reasons, including:
  • Lower total cost of ownership
  • Less strain on personnel
  • Reduced investment in infrastructure

Conclusion

MPLS adoption is again on the rise. Because of its low latency in data transfer rates, MPLS is exclusively suitable as an IP data transfer protocol enabling emerging productivity-boosting convergence applications in the organization such as:

• PSTN SIP trunking
• Native IP Multicasting
• IP videoconferencing
• Site-to-site WAN VoIP
• IP/PBX desktop VoIP

It should come as no surprise that MPLS is now quickly being adopted by the 30 percent of Canadian enterprise-level organizations that have not already made the switch. There are many drivers for this adoption, including its simplified configuration and ease of both connectivity and scalability. MPLS also promotes efficient bandwidth use while ensuring end-to-end quality of service. In these and other ways, the technology saves on scant resources. And because MPLS is a managed solution, IT personnel are able to spend less time on network administration and focus their attention on core issues.

There are clear best practices that should be followed in mapping QoS needs and in migrating to an MPLS network. The benefits of migration are clear and compelling, and there is little doubt that they will continue to accrue with the continued development of applications that break down the traditional barriers between telephony, video and data services. While MPLS technology may continue to mature slowly, it is likely to remain the gold standard data transfer protocol for many years to come. Where we are likely to see exciting new development is in the area of MPLS-enabled application acceleration networking.

For more information

Bell is a leader in IP VPN implementations and a pioneer in the field of MPLS. For more information visit bell.ca/entreprise, or request to be contacted by a Bell representative at bell.ca/contact-enterprise.

About the author

As a leader within the Bell Business Marketing and Product Development team, Greg Banks is responsible for the successful development and delivery of new product solutions for Bell business clients. Greg has worked for Bell since 1987, and has managed Customer Systems Engineering and IP Product Management teams for the past 12 years.