As we have moved away from data centers and offices to the cloud, SaaS, and the hybrid workforce, the overwhelming sentiment has been that enterprises and the enterprise network no longer need multi-protocol label switching (MPLS).

Is that true? Are we at the point where direct internet service and connection are all we need?

The short answer is yes…and no. Okay, it isn’t really a yes-or-no question. To answer this, we need to look at what MPLS is and what it provides.



MPLS providers have a vast global footprint with tons of connectivity. It has interconnected points of presence and local loops. The service providers have already built massive infrastructure, which means that an MPLS network is highly performant—fast, scalable, and secure.

Plus, compared to the public internet, latency and packet loss are more reliable and significantly lower. And to support, route-by-route assurances of availability (99.99% uptime per year), packet loss (.1% is usual), and latency are included with MPLS services.

Additionally crucial, MPLS services are established business-oriented services. They also include integrated billing, end-to-end distribution, management, and SLAs.

And as a result, MPLS circuits can provide guaranteed last-mile delivery for applications that need it in real-time. This benefit may seem minor, but enterprises have many applications developed long ago and can’t recover from degraded network performance (latency, loss, jitter) or suboptimal paths. These apps will do just fine on the internet under perfect conditions. What they struggle with is the imperfect nature that defines the internet.

There’s no way an enterprise can control the underlying path that traffic takes on the internet. One day traffic may go directly between New York and Houston, but the next day it routes through Michigan and Wisconsin before reaching its destination in Houston. That’s the way the internet has evolved. It’s a consumer-centric content delivery network. It accounts for the imperfect nature of the natural world and delivers content.

The guarantee MPLS provides compensates for the fact that these applications haven’t been (and likely won’t ever be) rewritten to perform well with the wild nature of the internet. That’s why we can’t use the internet in many cases.

Don’t get me wrong—the internet is great for many things. It’s great for content, high bandwidth, and agility. But the internet gives you no guarantees. So, it isn’t great for applications that need predictability.

It turns out that with its global footprint, private bandwidth, predictable paths, and functionality, MPLS is much better for applications.

It is a challenging problem. We need the incredible performance of MPLS, but we long for an MPLS Alternative. An alternative, more cost-effective model with increased agility of the internet. The root of the issue is that the way MPLS provides this enterprise-grade performance is precisely what causes it to be expensive and complex. Service providers had to build an expensive, monolithic private network to deliver MPLS.

So, back to our question: Will we need MPLS? Well, kind of. We definitely need the performance of MPLS but not the cost or the complexity.

What enterprises really need is a new type of network that combines the performance of MPLS with the lower cost, simplicity, and agility of the internet. See how the Graphiant Network Edge precisely provides this.