Return to the office: is your network ready for the influx of Zoom calls?

Remember enterprise networks before 2020? Before many enterprises around the world had to scramble to support an almost instantaneous switch to remote work? A lot has changed since then, and as more businesses start looking at a return to office-based work, or some form of hybrid working model they need to ensure their network is ready for the application mix that has become the new normal in the last 18 months.

Not that many years ago, I worked with several large enterprises on network refresh projects. Bandwidth was carefully planned, especially MPLS circuits that were treated like a precious commodity. Power users could have, possibly, 300-400kbps of bandwidth each. Everyone else? 100-150kbps per user. Voice traffic was calculated precisely based on the channels required, and protected using Class of Service on the MPLS backbone. And as for video? This was limited to specific conference rooms that had the expensive equipment for site-to-site calls, or back to a bridge of some sort. The exact bandwidth for each call, based on the codec and resolution, plus the correct allowance for signaling traffic was calculated down to the last kilobit per second. "Desktop" video conferencing? That was one of the scariest concepts to an enterprise network manager at the time - the unpredictable nature of it, and the risk of it overwhelming the network was enough to have it blocked in many cases.

This might seem laughable now, but there are many enterprise networks out there today that were architected in this exact way. A lot has changed since these users were in the office, but perhaps nothing will have a greater impact than the astronomical growth of SaaS video - the reality is that many networks just aren't ready for it.

All those video calls become your problem if the user is in the office

In some ways, it's shocking how quickly video became the new normal from the earliest days of the COVID-19 pandemic. This chart shows some of the rapid growth we saw:

With this growth comes a huge increase in bandwidth requirements. All of the modern SaaS-based video platforms use adaptive bitrates so these amounts are not fixed, but it's very common for video calls to exceed 2Mbps in each direction with good-quality cameras and ideal network conditions.

For home-based users, this isn't usually a huge issue. There's a whole separate set of issues related to Wi-Fi quality and other endpoint-level performance, but the typical home broadband connection can accommodate a few of these calls without too much of an issue. The limiting factor is usually the upstream bandwidth - there are some interesting stats in this recent article on the topic. The other good news for the enterprise is that this traffic often doesn't touch the corporate network, especially with the growing use of split-tunneling and cloud-based web security services.

Unfortunately, all this changes when large numbers of users return to the office and need to communicate with their teams, clients, partners, and everyone else they've been collaborating with over video for more than a year. A related point - video calls between users in different companies are now taken for granted - not that long ago these types of 'federations' were heavily controlled and required whitelisting, usually enabled only for specific close partners or joint ventures. Now video calling is pretty much universal between businesses of all types.

Hybrid working patterns will contribute to the use of these services, as users in offices still need to communicate with their colleagues that are working remotely.

If we look at the new demand this traffic puts on the network, it's clear that enterprises with thin, inflexible private WANs will struggle to achieve an outcome that delivers anything close to a good user experience. A different architecture is required to allow access to high-bandwidth SaaS applications like video, and to deal with the more dynamic nature of traffic flows overall in the network.

Building a more sustainable network with SD-WAN and SASE

The enterprises least likely to be impacted by this significant rise in the use of video are the ones that have already moved to an Internet-centric WAN with the ability to pool together multiple broadband circuits and effectively steer traffic in the optimum way. This has been possible for more than 5 years with high-quality SD-WAN solutions, and we are now seeing a sharp increase in the number of enterprises moving away from traditional telco-based offerings like MPLS.

The SASE architecture that has become popular in the last 18 months takes this a stage further, by delivering many of the security functions that were previously appliance-based as cloud services. This achieves two main benefits in the context of this issue - first, it is very well suited to a hybrid, work-from-anywhere model where users are just as likely to be connecting from home as from a shared workspace or office. Second, it allows more of the traffic to be kept away from on-premises bottlenecks. Both of these architectural benefits help deliver a better end user experience, and put the enterprise in a better position to deal with variable traffic patterns.

Most Coevolve clients that have adopted our telco-indepedent SD-WAN and SASE solutions have seen their average branch bandwidth increase by 10-50x, and most use two or more circuits at each branch office. This puts them in a strong position to deal with even significant changes to usage patterns like an increase in SaaS video.

If you're wondering how to deal with the impact of this new traffic on your WAN, get in touch! Many of our projects have started as tactical bandwidth augmentation initiatives that can be implemented in weeks, which are then followed by a more structured transformation.