Latency is one of the banes of the network engineer’s existence. Bottlenecks can occur for a variety of reasons, ranging from misconfigurations to subpar infrastructure. Whatever the case, minimizing latency and packet loss while optimizing throughput is one of the top priorities of IT divisions in enterprises across the globe.
While resolving bottlenecks on a LAN can often be achieved by troubleshooting and updating infrastructure where appropriate, the challenges of WAN performance are a different beast because they require organizations to account for infrastructure they have no direct control over. As businesses become more and more global in nature and cloud workloads continue to account for more and more of the traffic sent to and from enterprise networks, the importance of getting optimal performance out of WAN infrastructure is highlighted even more.
This means that selecting the right WAN solution is of the utmost importance. Given the dispersed nature of modern IT workloads and the need to connect users with cloud apps, branch offices, corporate datacenters, and other resources on the public Internet, cloud-based SDWaaS (SD-WAN as a Service) offers unique advantages to enterprises looking to get optimal performance out of their WANs.
In this piece, we’ll dive into some of the causes of WAN bottlenecks, what technologies and techniques can be used to address them, and how to best approach WAN optimization given the demands of modern IT workloads.
Major WAN issues that impact performance
The root cause of a given bottleneck can vary, but in general there are two specific sets of challenges that plague many organizations that must achieve connectivity at a global scale, those related to the “last-mile” and those related to sending data over long distances.
The last-mile refers to the stretch of a connection that connects the end-user to the service provider. The quality of this connection can vary significantly across the globe and many nations have less reliable connections than those commonly found in the urban areas of North America and Europe. This could be due to dated infrastructure in remote locations or oversubscribed connections leading to throughput issues (e.g. if multiple businesses are sharing the same cable or ADSL link). While the overall infrastructure of the Internet is steadily improving, there are still huge pockets of the globe that have subpar connectivity. This means that latency-sensitive applications (e.g. videoconferencing, streaming, VoIP, etc.) perform with significant degradation of service, if at all, for users in these areas.
Global connectivity challenges
Not only are there inherent “speed of light” problems when sending data long distances (think New York to Tokyo for example), packet loss often becomes a major issue when sending data across oceans. TCP packets that are dropped must be resent, exacerbating latency issues. Internet peering between the continents isn’t as fast as it could be, and dated infrastructure and over utilized connections also contribute to the challenges associated with creating a high-performance global WAN.
The shortcomings of MPLS as a solution
In the past, when an organization needed a reliable, dedicated link between two locations, MPLS (Multiprotocol Label Switching) was a common solution. However, in modern IT, MPLS has struggled to keep up. The two biggest challenges with MPLS at scale are:
- MPLS bandwidth is prohibitively expensive.
- MPLS is NOT optimized for cloud-based workloads and environments with significant Internet-facing traffic.
Expensive and not optimized for the cloud, aren’t exactly music to the ears of the modern CIO. What is needed is an affordable, reliable, and cloud-capable solution. This is where SDWaaS comes in.
Addressing the challenges of the modern WAN with SDWaaS
One of the inherent benefits of SD-WAN over MPLS is SD-WAN’s ability to be connection medium agnostic and leverage a variety of transports (e.g. fiber, cable, ASDL, 4G, etc.). This both drives down the cost of bandwidth (Internet bandwidth is much cheaper than MPLS bandwidth) and improves performance for cloud-based workloads (by solving the “trombone routing” problem). While this addresses the cost and “cloud-ready” side of replacing MPLS to resolve global performance issues, it doesn’t directly address the aforementioned last-mile and global connectivity challenges. This is where SDWaaS truly shines.
Premium, cloud-based SDWaaS providers have a global, SLA-backed, Tier 1 ISP supported backbone with Points of Presence (PoPs) across the globe. This means that premium SDWaaS has a robust backbone to traverse and bypass many of the inherent issues with the infrastructure of the public Internet across the globe. Further, given that many cloud infrastructure providers (e.g. Azure, AWS, etc.) have a shared network/datacenter footprint with SDWaaS providers, additional gains in performance can be had by switching to SDWaaS.
What about those last mile issues? SDWaaS has you covered there too. In addition to leveraging multiple transport methods, users of premium cloud-based SDWaaS can benefit from Active/Active failover, application QoS, and Forward Error Correction (FEC) to help address the myriad of issues that can occur between an end user and an ISP. As an added bonus, a full security stack is baked in to premium SDWaaS meaning not only can you send your data securely, you can avoid backhauling data for auditing and adding unnecessary latency to your connections.
SDWaaS enables economic, high-performance, global WAN infrastructure
Until the entirety of the public Internet’s infrastructure is operating with enough throughput and reliability to support secure end to end connections for everyone (don’t hold your breath), enterprises will need to select WAN solutions that help them ensure their business isn’t slowed down by bottlenecks and other performance issues. To compete on a global scale, connectivity at a global scale is required. SDWaaS enables organizations to connect their WANs in a secure, economical, and high-performance manner without adding undue complexity or cost.