MPLS or Global Private Backbone Network – Which is the right choice?

Global private backbones: Why the Internet is not enough

Stable buildings need a solid foundation, and this also applies to corporate networks. If the underlying connection is unstable, no application service can meet the required availability and performance characteristics. All of this is a problem for traditional SD-WAN and global enterprises. Internet routing is far too unpredictable to serve as the foundation for global enterprise backbones. The long distances and poor Internet peering between Asia-Pacific, North America and Europe exacerbate latency. Infrastructure problems and overcrowding increase packet loss. China is another particular challenge. Regulations require Internet traffic from China to be controlled by a centralized firewall (also known as the “Great Firewall of China”). As a result, China’s global internet connection has even higher packet loss and latency than other countries. If you don’t want to use the Internet as the basis for a global SD-WAN, what are your options? The traditional answer is of course MPLS. But several technological improvements now offer another option, which we call Global Private Backbones. Are Global Private Backbones right for your business? Let’s find out.

The argument for MPLS

As a privately managed backbone with built-in quality of service (QoS), MPLS does not suffer from the unpredictability of the Internet. Yes, it’s old wisdom, but the strengths and weaknesses of MPLS should be considered when transitioning to your global network. MPLS services provide the predictability that the Internet lacks. Whatever hurdles exist for the backbone are managed by the provider company, i.e. their carrier. Packet loss and latency statistics are more consistent and much lower than the Internet. In addition, MPLS services offer guarantees for availability (99.99% per year), packet loss (typically 0.1%) and latency on a regional basis. Just as importantly, MPLS services are mature services designed for business. Apart from SLAs, they have integrated billing, end-to-end delivery and management.

High costs for MPLS affect more than just the budget

But like everything, this type of dedicated, managed infrastructure comes at a price. The dedicated capacity, minimal latency and repair time make MPLS services very expensive. Anyone who has bought MPLS bandwidth for their business and Internet DSL for their home knows the surreal experience of paying three or even ten times more per megabit for MPLS bandwidth.

The cost of MPLS bandwidth doesn’t just affect the bottom line. IT managers need to cut their bandwidth spend to stay within budgets. For this reason, branch offices are being equipped with connections that are just big enough. These narrow connections are increasingly incompatible with today’s larger data streams. With increased bandwidth requirements, companies often invest in additional equipment to make the most of MPLS. This is where WAN optimization becomes particularly important. The additional equipment increases capital costs, but also makes management and troubleshooting more difficult.

Internet and cloud traffic: exacerbating the MPLS problems

The commitment to dedicated capacity, minimal latency and repair time make MPLS services expensive. The problem is exacerbated when traffic moves to the internet and the cloud. Providing direct internet access to remote offices requires securing that connection with a whole host of advanced security services. To avoid these costs, many MPLS-based companies centralize Internet access. However, centralizing Internet access requires Internet and cloud-bound traffic to be routed back to the Internet portal. This creates network-atypical traffic. This consumes precious MPLS capacity and can degrade internet and cloud performance due to the well-known trombone effect.

Limited agility

Less well known, but perhaps just as important, is the inflexibility of MPLS services. New installations can take anywhere from 30 days to six months, depending on location and infrastructure. Since bandwidth is often limited, new applications or changes in application dynamics can force bandwidth upgrades that can take weeks as well. And you are tied to the network coverage of the respective telecommunications company. Inevitably, some locations are outside the carrier’s MPLS coverage area. Telecom carriers must connect their network to other local or regional telecom carriers, which increases costs. In short, MPLS services often no longer fit the focus of today’s digital transformation.

Depending on the location and existing infrastructure, they can take up to six months to deploy. But they are a necessary evil for companies, or so it seems. Changing dynamics always lead to innovations that allow us to replace old solutions with new ones. MPLS is no exception.

What speaks for SASE

Gartner has recently defined a new architecture, Secure Access Service Edge or SASE, that can overcome the limitations associated with technologies such as MPLS. SASE solutions are cloud services that connect and secure all areas of the enterprise such as data centers, branch offices, mobile applications and cloud resources. This is achieved through the convergence of network and security solutions into a seamless cloud service. The provision of a global private backbone is crucial for a SASE architecture that is able to replace MPLS, without the unpredictability and congestion of the public Internet.

As an alternative to MPLS, global private backbones provide a global, predictable and secure network backbone with MPLS-like latency at a lower cost. Typically, building global backbones has required significant infrastructure investments in purchasing or leasing capacity across wavelengths and optical equipment, MPLS switches, IP routing and more. The combination of these factors increases telecom companies’ build-out costs, costs that are ultimately paid by customers. Instead of creating the infrastructure for a global network, global private backbones are built as a geographically dispersed, SLA-backed network of PoPs interconnected by multiple Tier 1 carriers. There are none of the high costs of MPLS or the unpredictability and congestion of the public internet.

SASE combines SD-WAN with the backbone core for more flexibility at cloud level

SASE backbone offers better performance, availability and coverage than any single provider.

No single Tier 1 carrier can reach all parts of the world, nor can a single Tier 1 carrier provide the predictability of MPLS. Just as organizations use SD-WAN to aggregate Internet services to overcome the limitations of a single service, SASE uses SD-WAN to aggregate Tier 1 carriers to overcome the limitations of a single network. The SASE backbone consists of Points of Presence (PoPs) that connect to multiple carriers. The PoPs form a software-defined overlay across these networks and dynamically route packets based on real-time metrics, application requirements and corporate policies. As a result, the SASE backbone provides better performance, availability and coverage than any single provider.

Improve uptime and reduce costs with COTS hardware and distributed software

SASE backbones are designed to house the resources required to provide SASE functions, including servers, network connectivity and software. Each PoP is made up of multiple, redundant computing units without relying on proprietary devices for core PoP functions. In addition, the PoP core functions are implemented as fully distributed software. In the unlikely event that an entire PoP fails or becomes unavailable, traffic is seamlessly redirected to another PoP.

There is no direct link between a customer site or user and a specific PoP resource. As there are no major hardware dependencies, PoPs can be put into operation very quickly anywhere. All that is required is commercially available hardware (COTS). PoPs close to a customer’s location shortens the “last mile” and enables sophisticated routing and traffic control to optimize long distance traffic.

Edges are connected to the nearest PoP via the general Internet. The impact of the Internet is minimized by using enterprise Internet services and connecting to PoPs within 25 milliseconds. Last mile availability is improved by using differently routed connections. The result: uptime can far exceed typical Internet connectivity and dedicated Internet access, MPLS resiliency for local networks.

MPLS or affordable global private backbone?

Every major upheaval begins with a displacement of the “tried and tested”. The cloud displaced virtualization, which displaced the server industry. Backbones are no different. With SASE platforms, you have the same level of consistency and predictability as MPLS. It still gives you integrated, advanced security, ubiquitous coverage and support for cloud, SaaS and mobility at the same time, why would anyone pay ten times more for MPLS?