Controlling Video Conferencing Traffic with Doug Schaefer

How and why to approach a hybrid setup for managing video collaboration in Azure/AWS/GCP, differences from full SaaS out of the box, and manipulating call signaling behaviors. A niche UC topic, but also an underappreciated one. Doug Schaefer, CTO at the value-added reseller American Sound and a first-time Cloud Field Day presenter, framed his talk around solving real-world problems rather than pitching a product. He described how the pandemic pushed video conferencing from an afterthought to a business-critical service, and how enterprises now routinely run several platforms at once. He explained that in SaaS platforms, where a meeting’s conference bridge is hosted depends on the platform, the meeting host, and the region of the host’s tenant, and that these rules differ enough between vendors that network design has to account for them. His first case study was a global investment bank with two to three thousand Teams and Webex rooms whose international calls kept falling apart. The cause was proxy routing: a New York room’s traffic passed through data centers in the Mid-Atlantic and Southeast before reaching an Azure Front Door, adding roughly 170 to 200 milliseconds of latency. That was enough to degrade calls once participants in other regions joined via different paths.

Schaefer then contrasted the free, WebRTC-based Direct Guest Join interop that comes with Teams, Zoom, and Google Meet against SIP-based interop. SIP delivers higher-definition video and lets organizations insert their own infrastructure to anchor call legs. He outlined the available architecture models: fully on-premises deployments, where Cisco Meeting Server/Expressway and Pexip Infinity are essentially the only two options; hybrid setups; and full vendor SaaS, which is the most common. In the hybrid “Cloud Video Interop” approach, conference room codecs connect to a media node the organization hosts in Azure, AWS, or even Oracle Cloud. That node routes the call to the appropriate Teams or Zoom connector in the region, giving every platform the same predictable network path. His second case study was a healthcare system that moved its video infrastructure into Azure to meet recording, auditing, data sovereignty, and Epic integration requirements. Telehealth Zoom calls scheduled through Epic were routed through the system’s own Pexip nodes, and only approved call types could reach third-party services. In response to an audience question, he said security and performance improved together, since the customer could provision far more bandwidth to Azure. The real trade-off was cost, which required careful total-cost-of-ownership modeling compared with simply buying licenses.

Comparing network paths, Schaefer noted that native SaaS media is out of the organization’s control as soon as it leaves the client, which also makes troubleshooting harder. The interop path, by contrast, keeps SIP traffic visible and manageable. It can travel over network-as-a-service providers such as Cloudflare or Megaport, or over dedicated links like Azure ExpressRoute and its AWS, Google, and Oracle equivalents, to a self-managed media server and then directly into Microsoft 365. To address the business case, he cited relatively independent research, including work from Owl Labs. It shows an average of about six minutes lost per meeting to startup problems and one to three hours of IT time per week spent troubleshooting video systems. He closed by acknowledging that many organizations’ current setups may be adequate. However, he argued that as vendors add AI features and shared collaboration content to meetings, knowing exactly how media travels from point A to point B will matter more for security and troubleshooting. Hybrid architectures, he said, give architects a way to offer customers that control.