MP-TCP link bonding protocol offers declining MPLS a much needed life-line

Recent work on a new approach to Hybrid Access emerging from one of the industry standards body is likely to be music to the ears of the major carriers who have seen a steady erosion of their market share by the more agile Internet service providers.

Despite the many benefits that an MPLS based VPN connection can offer businesses, particularly in terms of security and SLA guarantees, the major carriers have struggled to prevent customers opting to move some or all of their WAN architecture onto a low-cost, high-speed Internet link and VPN as soon as contracts allow.

To some extent this trend is an understandable consequence of the growth in the uptake of Cloud applications as the basis of corporate communications. Most network managers agree that MPLS is not ideally suited to handling large volumes of traffic and has led to network congestion and performance headaches for IT teams, for which just increasing bandwidth is not necessarily the solution. In addition there are still lots of businesses that choose a hybrid approach to hosting applications with some in the corporate data center and others in the Cloud. This means that moving to an all Internet infrastructure is not going to solve the problem in all cases either.

The obvious answer is to opt for some form of hybrid MPLS/Internet access ecosystem that can provide granular control of all traffic across the WAN and ensure that load can be spread across multiple links based on a range of business priorities and policies. While this approach is great in theory the reality can mean high capex investment in load balancing for a sub-optimal solution.

Application-aware policy-based traffic distribution means any given application/session is restricted to a single link’s bandwidth, which leads to the expensive MPLS being under-utilized without adding any link failure reliability. With cost-saving also a major factor for businesses looking for an alternative to MPLS it is easy to see why some companies decide to go for the pure-play Internet-based option, even at the cost of losing the reputed SLA and security benefits offered by MPLS.
However, recent exciting developments emerging from the Broadband Forum are promising to enable advanced hybrid access functionality to be embedded in the CPE, which is great news for carriers looking to be able to offer customers the type of services they need while retaining their private corporate VPN at an acceptable price-point.

While different CPE-based hybrid access flavours have been available from vendors like ourselves for some time, the MP-TCP protocol being considered by the BBF is taking the concept to a whole new level in terms of policy based traffic distribution. This breakthrough, derived from the Multi-Path TCP protocol, enables delivery of innovative Hybrid VPN business models employing layer 4, multi-link load-balancing paradigms that are far more flexible and powerful than session-based application policy solutions.

Key features of the new MP-TCP protocol include support for packet-based flow distribution. This means application/session traffic can be sent on all links according to their characteristics so that any given application/session may use the aggregated bandwidth from all the available links, with no service interruption if a link fails.

OneAccess has been at the forefront of innovation for multi-link solutions that efficiently bond dissimilar link access technologies, such as xDSL, MPLS, LTE or satellite. As an active member of BBF, we have been a major contributor to the MP-TCP based solution, which has given us a head start on developing powerful traffic distribution solutions for service providers as a platform for their Hybrid VPN services.

Recognizing the constantly changing performance characteristics of individual public links that can be affected by shared usage factors such as a peak traffic period on the LTE mobile network, OneAccess’ dynamic link estimation feature has been added to optimize traffic distribution. This means that distribution decisions are based on the actual capacity of the individual links in terms of the optimum speed, latency and bandwidth appropriate to each class of traffic. It also means that packets in the same application/session can be distributed across multiple paths to leverage MPLS bandwidth capacity for better performance and reliability before spilling over to the other VPN access link(s).

Available as a software module on a physical CPE or as a VM (or soon as a VNF), carriers now have a range of proven off-the-shelf solutions that can help extend the life of their MPLS infrastructure investments and offer a better range of choices for customers who want the best of both worlds.