Head office (South Africa)

​The Campus, 57 Sloane Street
Cnr Sloane Street and Main Road
Bryanston, Johannesburg, 2021
+27 11 575 0000

Technical support

+27 11 575 2571 / 0860 221 221
+61 3 9626 0497 / 1800 638 457
+55 11 3878 6500
+31 342 402 859
North America
800 974 6584
United Kingdom
+44 1252 779 779
​Are hybrid wide area networks the nail in the coffin for MPLS?

​Are hybrid wide area networks the nail in the coffin for MPLS?


Are hybrid wide area networks the nail in the coffin for MPLS? 

A maturing Internet and the increasing adoption of cloud computing are changing the face of corporate data management and delivery for good. 

Organisations are gradually moving to a more dispersed data centre model because not all workloads need to be hosted in a single, centralised facility anymore. Data sets can now be divided among owned or hosted data centres across a wide geography, as well as across a number of different cloud providers situated anywhere in the world. 
​​'Working together as a hybrid WAN, these alternatives may not only save on connectivity costs, but also provide greater capacity and perfomance,'​​
Previously, connectivity between branch offices and a centralised data centre was, in most cases, provided by a dedicated, private wide area network (WAN). In the great majority of cases, this network was based on multiprotocol label switching (MPLS) technology offered by the larger tier-1 telcos. 

MPLS networks are still popular today, however, as organisations increasingly use dispersed data centres, they’re also realising that several other connectivity alternatives may make sense from a cost and efficiency standpoint. Working together as a hybrid WAN, these alternatives may not only save on connectivity costs, but also provide greater capacity and performance. So does this spell the beginning of the end for MPLS as we know it?

Adding up the numbers
One of the main problems that a dispersed data centre model causes for the traditional MPLS WAN is weakened application performance. ‘For example,’ explains Raoul Tecala, Business Development Director at Dimension Data, ‘there’s the issue of “tromboning”. This occurs when users at a branch office access their cloud-hosted applications via the corporate MPLS network to the organisation’s central data centre, and then out over the Internet to the cloud provider. To fulfil the request, data would then have to flow all the way back to the user along the same route. Depending on the physical location of the cloud service provider, potentially thousands of miles could be added to the round-trip time, which could cause application performance degradation. Applications that run on dispersed data centre infrastructures can therefore become slow and cumbersome to use.’  

Tecala points out that organisations will not only solve application performance issues, but also save costs by enabling application data to reach the user via more than just an MPLS network. ‘In your average branch office today, there may be a 10-megabit link provided by a telco, which usually costs in the region of USD 900 per month. But organisations can now supplement the MPLS network with lower-cost bandwidth alternatives, such as the public Internet. They can, for example, get approximately 100 megabit direct Internet access for around USD 150 per month. And this is just one of the possible options. There are also Metro Ethernet and 4G LTE to consider – each more cost-effective than MPLS.

‘By offloading traffic destined for the Internet to lower cost alternatives, organisations will realise that they don’t need as much MPLS capacity as before and perhaps reduce the 10-megabit access to a 5-megabit alternative which will cost them approximately USD 450 per month. So for connectivity to one branch office, the business previously paid USD 900 per month for 10 megabits, but through the use of less expensive connectivity alternatives it now pays only USD 600 per month for a combined 105 megabits. That’s over ten times more capacity at a 33% reduction in cost.’ 

What about performance?
Cost savings aren’t the only benefits that hybrid WANs would have. ‘They may also solve application performance issues,’ says Tecala, ‘by enabling the routing of each application’s traffic via the channel with the best probability of delivering the necessary quality of service. In addition, organisations may improve overall network availability thanks to the real-time selection of diverse routes, where each link can serve as a backup for the other.

‘Currently, there are differences between business-class direct Internet access and an MPLS network. For example, you can’t get performance guarantees on a basic Internet connection. Since an MPLS network is dedicated and private, service providers can provide performance service level agreements on aspects such as latency, jitter, and other technical factors. The question organisations need to ask themselves is: What is the correct type of connection for each application? Given the massive difference in price, application traffic that truly requires the high performance should traverse MPLS, while alternative, less expensive connections can be used for other traffic types. 

‘Keep in mind that the Internet has matured significantly since it first emerged 15 years ago. At that time, it couldn’t provide reliable business-class connectivity. However, over time, it has improved significantly,’ says Tecala. 

So, if an organisation can get better application performance through a hybrid WAN strategy, why have an MPLS network at all? Is the writing on the wall for MPLS simply because the economics of it don’t make sense anymore? And if more organisations choose multiple connectivity options delivered by multiple service providers, will there still be any use for an MPLS network in ten years’ time?
Says Tecala: ‘While there will always be a place for MPLS-type services, the trend is clearly toward a multiconnectivity strategy. Over the next five to 10 years, I think there’s a strong chance that the revenue from traditional MPLS services will be a fraction of what they are today – both in terms of a reduction in demand, as well as reduced prices from MPLS providers. An analogy here is voice traffic – a generation ago, service providers made a significant revenue from carrying voice traffic where it was common to see toll rates in the dollar-per-minute range. Now however, voice is seen as just another application to run over a network, such that the amount of voice minutes on carrier networks have dropped significantly, as have the rates for voice traffic one down to cents per minute. Ten years from now, MPLS may go the same way.’ 

​​'Is the writing on the wall for MPLS simply because the economics of it don't make sense anymore?'​

Slowing the process
Despite the obvious benefits, Tecala explains that there are still factors holding organisations back from the immediate adoption of a more cost-effective hybrid WAN strategy. ‘Most businesses have signed extended MPLS contracts with their respective service providers. These may be for as long as three to five years, which ties them to using the MPLS in the short to medium term. Adding alternatives to the connectivity mix would only add to their telco cost.’

Another factor that may slow down adoption is the highly publicised Internet ‘brown-out’ failures. Says Tecala: ‘This may be caused by, for example, a large peering point that malfunctions and thereby disables a significant portion of the world’s connectivity. Ultimately, it doesn’t really matter whether this is a real threat or not – it’s the perception that counts, because it may strengthen the idea that the Internet isn’t trustworthy. It doesn’t even have to be an actual outage; it could also be a highly publicised security breach such as a major hacking incident that causes panic and suspicion about the Internet’s integrity.
Ironically, there’s a perception in the market that MPLS networks are more secure than connectivity services provided over the public Internet which, in reality, is not the case. But in choosing your enterprise WAN connectivity, perception is reality. 

‘The third delaying factor,’ says Tecala, ‘might be price pressure. Currently, telcos charge their expensive MPLS rates because there’s no real competitive threat yet. The concept of multiconnectivity based hybrid WAN is still in its infancy. But once telcos start seeing more competition from hybrid alternatives, they’ll have no choice but to reduce their MPLS rates in order to stay in the game. This means that their MPLS revenues may drop, but probably won’t disappear altogether. 

‘The overall conclusion is that MPLS probably won’t die out completely, but it will certainly need to be far more cost-competitive in order to survive. MPLS will probably also become just one of several ways in which organisations can connect, rather than the only or most important way.’ 

Knowing what’s on your network 
To take advantage of the cost benefits offered by the Internet and other alternative connectivity models, Tecala advises to start with an understanding of the applications you have running over your network, and what traffic volumes and patterns they create in terms of their peaks and troughs. With this application traffic information, you can design a hybrid network that utilises the types and amounts of connectivity that’s right for your environment. ‘Your network can then be designed in such a way as to detect when there’s any form of performance degradation, and to optimally reroute the traffic via another channel. 

‘A hybrid WAN enables you to take a real-time, application-centric approach to network traffic management. But you also need to design the environment more carefully by picking the right primary and backup recovery routes, so that you’re able to use the network effectively and with confidence,’ concludes Tecala.
​'The overall conclusion is that MPLS probably won't die out completely, but it will certainly need to be more cost-competitive in order to survive.  MPLS will probably also become just one of several ways in which organisations can connect, rather than the only or most important way.'​​