The WAN Accelerator and Modern Network Optimization

May 5, 2020

Network latency costs money. This is a simple concept most IT professionals understand. However, when I discuss latency reduction and WAN acceleration with network managers and CIOs, one of the key takeaways is that getting network optimization right has changed significantly over the last decade. While WAN optimization and acceleration are still important, increased bandwidth availability, cloud, and mobile have significantly shifted the paradigm. So, what exactly are WAN accelerators and what is WAN acceleration in 2020? Here, we’ll answer those questions.

What is a WAN accelerator

Simply put, a WAN accelerator is any hardware or software appliance that provides bandwidth optimization across a WAN. There are a variety of different techniques that different WAN accelerators, also known as WAN Optimization Controllers (WOCs) use, and these include:

  • Compression that reduces the amount of data sent across the network. Compression, in the context of WAN acceleration, typically operates at the byte-level and works in a similar fashion to file compression but applies to data in transit.
  • Deduplication is similar to compression but operates on larger amounts of data, typically at the block level. Its goal, like compression, is to maximize the available bandwidth.
  • Caching is another technique focused on reducing bandwidth usage. Caching stores frequently accessed data locally, eliminating the need to retransmit the data across the network.
  • Protocol acceleration techniques improve protocol operation across the network, particularly in terms of reducing the latency introduced by inefficient protocol operation. Local flow control, selective acknowledgment, and window scaling are techniques that help enhance TCP connections.
  • Application-specific acceleration techniques boost the efficiency of applications. While protocol acceleration improves the operation of the underlying network and specifically the TCP-layer, application-specific optimizations address the chattiness of application-layer protocols.
  • Packet loss correction techniques, such as packet duplication, for overcoming packet loss particularly in the last mile.

Generally, WAN acceleration appliances were deployed at locations across a WAN to achieve WAN optimization objectives.

SD-WAN: The WAN accelerators for the modern digital enterprise?

As we can see, in the past WAN acceleration was heavily focused on reducing bandwidth consumption between sites. This made sense when applications resided in the private datacenters and were accessed from branch offices across narrow, expensive MPLS circuits. However, today, applications and data have shifted to the cloud and accessed as much by mobile and remote users as those in the office, rendering appliances obsolete. And with Internet capacity far more readily available and more affordable than MPLS, conserving bandwidth is no longer nearly as critical.

What is necessary is the ability to leverage Internet capacity in a way that can meet enterprise requirements. SD-WAN edge appliances run affordable, last mile public Internet services in active/active configuration. Not only does this give companies incredible agility in combining bandwidth capacity but also adds last mile resilience. In the event of a brownout or blackout, SD-WAN devices can switch traffic to the alternate service. And by including packet loss correction techniques, particularly packet duplication, SD-WAN devices can overcome last-mile connectivity problems.

At the same time, edge-based SD-WAN continues to fall victim to the same limitations as any appliance. The short history of SD-WAN shows that an appliance-based approach works for site-to-site connections but continues to be a poor fit for the cloud and irrelevant to mobile devices. Additionally, the shift from MPLS to a public-Internet core, on which edge-based SD-WAN depends for its cost savings, introduces a myriad of challenges endemic to the modern Internet infrastructure that can negatively impact the performance of latency-sensitive applications, such as VoIP (Voice over IP) and UCaaS (Unified Communications as a Service).

This creates a situation where the modern digital enterprise needs an approach to WAN optimization that keeps bandwidth costs low, resolves the reliability and latency challenges of the public Internet, and accounts for cloud & mobile use cases. The cloud-native approach to WAN optimization directly addresses all of these challenges.

The cloud-native approach to WAN acceleration

Instead of hosting WAN acceleration in appliances at edge, the capabilities are increasingly being moved into the cloud. Making WAN acceleration part of a global, cloud-native platform, like Cato Cloud, eliminates the appliance form-factor that was so difficult to deploy in the cloud and irrelevant to mobile users. Instead, Cato and other cloud-native platforms let organizations use the optimum solution to connect their “edges” — a simple SD-WAN device for sites, native cloud connectivity for cloud resources, and client-based or clientless connectivity for mobile and remote users.

Regardless traffic is sent to the nearest PoP where the cloud-native software accelerates traffic and delivers it across the Cato backbone to the respective edge. The PoPs of Cato Cloud are collocated in the same physical datacenters as the IXPs of the leading cloud datacenter providers. With a few clicks on a management console, cloud traffic can be sent across Cato’s accelerated backbone and dropped at the footstep of the cloud datacenter provider or at the PoP closest to the cloud application provider. Additionally, by segmenting connections in a last-mile, middle-mile (a global private backbone), last-mile paradigm Cato Cloud is able to recover from packet loss faster than SD-WAN appliances.

As a result, Cato Cloud users benefit from:

  • Optimized global connectivity. Cato’s global private backbone consists of 50+ PoPs supported by multiple Tier-1 Internet Service Providers and is backed by a 99.999% uptime SLA. This helps enterprises address the reliability and performance challenges of the public Internet across the middle mile without sacrificing flexibility for cloud and mobile applications.
  • Network Optimization. Cato boosts end-to-end throughput by minimizing the effects of latency on traffic flow. Bandwidth-heavy tasks such as file uploads and downloads can improve by 20x or more.
  • Cloud application acceleration. Cato routes traffic from cloud applications, such as UCaaS and Office 365, along the optimum path to the PoP closest to the customer’s instance in the cloud. Traffic is dropped off at the doorstep of the cloud application provider. In this way, Cato minimizes latency in cloud application sessions and by applying its WAN optimizations, further reduces the effects of latency.
  • Cloud acceleration and control. Cato routes traffic from all WAN edges to the Cato Point of Presence (PoP) nearest to the cloud service provider’s datacenter. As Cato shares a datacenter footprint with many popular cloud service providers, latency from the Cato PoP to the provider is near zero. Further, Cato provides this functionality without the need for cloud appliances and without the additional cost of services such as AWS Direct Connect or Azure ExpressRoute.
  • Mobile access optimization. Using clientless browser access with mobile or with the Cato Client application, enterprises eliminate the need for inefficient backhauling and remote users automatically connect to the closest Cato PoP and receive the same enterprise-grade optimization and protection as on-premises users.

Just how much of a difference can Cato Cloud make in the real world? Looking at Salcomp’s experience Cato Cloud was able to provide a better than 40x throughput for Sharepoint file transfers.

Modern WAN acceleration requires a modern approach

WOCs were built to solve a specific set of problems that existed when bandwidth costs and availability were the primary WAN acceleration and optimization challenges. Today, cloud and mobile use cases coupled with reduced bandwidth costs have changed how enterprises need to approach optimization. Cato Cloud offers enterprises an approach to acceleration made for the digital business, one that optimizes traffic of all tenants of the new enterprise, not just locations. If you’d like to learn more about what Cato can do for you, contact us today or start a trial to put Cato Cloud to the test.

 

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.