Why Global SD-WAN powered by IP Transit Backbone is Perfect for the Post-MPLS Era

August 17, 2017

The Search for Affordable MPLS Alternative

Global organizations are looking for SD-WAN services to provide an affordable, MPLS alternative. If you are already using MPLS, you are well aware of its challenges: high costs, rigidity, long time to deploy and incompatibility with the growing demand for direct cloud and internet access. For a long time, organizations made a tough tradeoff: pay the price of MPLS for a consistent and predictable network experience, or use the affordable but unpredictable, best-effort public Internet. Now, a new breed of cloud networks is leveraging technological and business advancements in global connectivity to create a high-quality alternative: a network that is more affordable than MPLS and more consistent than the public Internet.

We discussed at length what makes the public Internet a problematic WAN backbone. At a very high level, the Internet is not orchestrated to ensure global routing is continuously optimized for minimal Round Trip Time (RTT) through optimal route selection and packet loss mitigation. There are two main reasons why Internet orchestration isn’t possible. First, the service providers that comprise the Internet are making routing decisions based on commercial interests, and not optimal performance. Second, the protocol that binds the Internet together, BGP, is not built to consider the changing conditions of Internet routes, such as packet loss, latency and jitter, just the distance, measured in hops, between source and destination.

To build a credible alternative to MPLS, we need to address global orchestration that enables dynamic routing based on end-to-end route quality. But, in order not end up with the same cost and complexity as MPLS, we need to use a lower-cost platform. This is where IP transit enters the picture.

IP Transit and the Public Internet

IP transit is a global connectivity approach used by providers such as NTT, PCCW, Telia and GTT. These providers deployed global backbones with huge capacity that carries the majority of Internet traffic today. No single carrier covers the entire globe, but each has substantial intercontinental footprint. The network buildout of the last decade created excess capacity, which drove drown the price per megabit. Capacity was added to accommodate the growth of global platforms, such as Facebook, and Amazon, as well as increased traffic from DDoS attacks. This additional capacity has created a low-cost backbone option for businesses with SLAs on global round-trip times. However, complex contracting and volume-based pricing makes IP transit only accessible to the largest of enterprises.

The public Internet simplifies enterprise networking through local ISPs that use Internet exchange points. IXs enable regional and global ISPs to share Internet routes and capacity with each other through Internet peering. However, Internet exchanges still mean enterprises can’t control which ISP carries their traffic, and how routes are selected.

Cato Cloud: Globally orchestrated, Quality-aware Overlay of IP transit providers

Cato has built a cloud network that leverages the low costs, high capacities and SLAs offered by IP transit providers. Cato uses advanced software to dynamically optimize global routing over multiple IP transit providers. As with any cloud service, Cato unburdens enterprise IT by assuming the complexity of contracting, deploying and orchestrating this global network. And, by using commodity hardware and its own software, Cato can pass the aggregate benefits of IP transit to customers in the form of very competitive pricing.

How does it work?

Cato has built a global network of PoPs from infrastructure in tier-1 datacenters and global cloud providers. Cato directly contracted with multiple tier-1 IP transit providers and bought massive SLA-backed capacity. In places where we can’t directly access tier-1, IP transit providers, we rely on Amazon AWS and other leading cloud providers who also use IP transit services from tier-1 providers.

Cato’s PoP software leverages the global underlying providers to create a fully meshed, tunnel overlay between all PoPs. The PoP software continuously measures route quality, tracking statistics such as latency and packet loss rates. The impact on RTT is minimized through WAN optimization techniques.

Using this cloud network architecture we achieve global orchestration of traffic routing by dynamically selecting the best route between customer locations. Cato monitors the packet loss and latency of all relevant routes, and continuously identify the best route, based on RTT, in real time.

The benefits of the above architecture are twofold. First, the Cato Cloud provides better and more consistent RTT than the public Internet in most global scenarios, approaching MPLS numbers at a fraction of the cost. This helps support global deployments of latency-sensitive applications, such as voice and Remote Desktop (RDP). Second, because the Cato Cloud optimizes the last mile independent of the middle mile, it can maximize global throughput for bandwidth-intensive applications, such as backup and file transfer. Throughput is maximized by  maximizing TCP window size which is made possible by reducing the time to detect and recover from packet loss.

Cato provides organizations of all sizes with an affordable, global backbone

Cato isn’t “using the Internet” in its unmanaged and uncontrolled form. Cato leverages the Internet for what it’s best at — access. The Cato backbone, though, is built from a combination of sophisticated software, commodity of the shelf hardware, and affordable, high capacity, SLA-backed IP transit infrastructure. It’s a modern architecture that can support enterprise connectivity requirements for branch locations, cloud applications and infrastructure, and mobile users anywhere in the world.

Could Cato use a more expensive, dedicated transport for its cloud network? Sure. But, customers would have to pay a hefty price for no real improvement in service. In our view, if it costs like MPLS and behaves like MPLS, it is basically MPLS. We chose to innovate and use abundant and affordable global capacity, coupled with our proprietary software to dramatically reduce the price of connectivity, deliver consistent and excellent service and make it available to organizations of all sizes.

Dave Greenfield

Dave Greenfield

Dave Greenfield is a veteran of IT industry. He’s spent more than 20 years as an award-winning journalist and independent technology consultant. Today, he serves as a secure networking evangelist for Cato Networks.