Network Computing is part of the Informa Tech Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them. Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.

From MPLS to the Internet: Optimizing Enterprise Networking in a New Era of WAN

(Source: Pixabay)

The WAN looks different than it did a decade ago. This much is obvious.

When I think of benchmarking enterprise wide area network costs throughout the 2010s, I think of customers who wanted to ensure that their carrier—often a single global telco—had lowered MPLS prices along with the market over the life of their multi-year contract.

But this is no longer the blueprint for WAN benchmarking.

Today, many large enterprises have a WAN that includes multiple underlay technologies from several vendors. Cloud apps and services from Infrastructure as a Service (IaaS), Software as a Service (SaaS), and Unified Communications as a Service (UCaaS) are the name of the game.

As I see it, there are four major forces influencing the WAN's evolution away from MPLS and toward the Internet.

Data centers are migrating

Here’s what we know. The corporate data center has been migrating for several years.

TeleGeography’s 2019 WAN Manager Survey of 64 large multinational enterprises found that, while about one in 10 still had all their data centers on-premises, nearly one in five had already moved off-campus.

The majority had a mix of on-premises and off-premises configurations. But the upshot is that most corporate IT infrastructure teams saw a shift in traffic patterns away from hubs connected to their MPLS-based intranets.

Cloud delivery models

Most enterprises are now also multi-cloud. Nearly three in four WAN Manager Survey respondents had two or more IaaS partners.

Further, those enterprises were connecting to their IaaS providers using multiple methods. While some kind of dedicated interconnection was most popular, more than half were using IPsec VPN; two in five included basic internet connections. This is beyond the explosion of SaaS/UCaaS products in the enterprise that often originated as consumer products delivered over best-efforts broadband connections.

It makes little sense to push all this internet-native traffic through expensive MPLS circuits to central breakouts. So, there was pressure to shift traffic to local internet breakouts even before enterprises began adopting software-defined wide-area networks (SD-WAN).

SD-WAN adoption

In 2018, fewer than one in five enterprises had adopted SD-WAN.

By 2020 that had doubled to more than two in five. One in four respondents were in the pilot or rollout phase.

And why are these enterprises adopting SD-WAN? The top two reasons cited were the need to boost site capacity and select alternative products for the underlay.

Lower Internet Costs

There’s no doubt that the difference in price between MPLS and internet alternatives also contributes to enterprises’ desire to increase capacity and use alternate products. Our survey data indicates that MPLS port sizes skew lower than the Internet.

Its enterprise services pricing research also indicates there’s still a gap between DIA and MPLS port pricing, let alone the fact that DIA is often available without additional access line charges while MPLS is not.

What’s a WAN manager to do in a post-MPLS world?

This all points to a clear trend – enterprises are relying less on MPLS and more on the Internet.

In 2018, 82% of our WAN Manager Survey respondents’ sites were running MPLS. By 2020 that had fallen to 58%. DIA and broadband gained ground over the same period. These factors have pushed the corporate WAN away from MPLS and toward the Internet.

But what are the implications for WAN managers and their vendors?

Firstly, the WAN manager must consider the geographic footprints of their cloud service providers (CSP), not just their own data center sites. Staying on top of CSP footprint expansions is huge.

Enterprises must also select data center locations from literally thousands of sites operated by hundreds of different providers. Data centers differ not just in costs for power, cross-connects, and rack space but by CSP on-ramps, network provider presence, and the ability to reach exchanges or fabric providers. There's simply more to consider.

This brings me to the last novel concern. WAN managers must appraise what happens to their traffic once it leaves their office over a best-efforts internet connection. Performance on the “internet middle mile,” once the telco’s problem in the MPLS network, is now a concern for the enterprise.

Whether it’s selecting a tier 1 Internet Service Provider (ISP), using a Network as a Service (NaaS) provider, an Over-The-Top (OTT) middle mile optimization tool, or joining peering exchanges, WAN managers now have entirely new vendors or solutions to evaluate to ensure end-to-end performance.

The key now is how WAN vendors—whether telcos, Systems Integrators (SIs), Managed Service Providers (MSPs), or novel solutions—are able to work with and navigate these changes alongside the multinational enterprise.

Greg Bryan is the Senior Manager of Enterprise Research at TeleGeography.