A head office uses Cloud OnRamp for SaaS for Microsoft 365 with DIA at three branches and a gateway site. Users complain about Teams quality when a branch Internet path degrades. Which Cloud OnRamp for SaaS behavior best selects the best exit for the SaaS application?

The correct behavior uses Cloud OnRamp for SaaS with continuous probing of Microsoft 365 front doors to inform per-site, per-application exit selection between local DIA and the gateway site, which directly resolves Teams quality issues when a branch internet path degrades because the affected site can automatically use the better performing exit without manual changes. The gateway-preferred centralization approach fails by concept because pinning Microsoft 365 breakout through the gateway for inspection removes dynamic best-exit choice and forces hairpinning even when local DIA is healthy or the gateway path is degraded, adding latency for real-time media. Overlay Application-Aware Routing with SLA-based tunnel selection fails by concept because it optimizes the SD-WAN overlay path using BFD and transport SLAs rather than measuring SaaS cloud performance from each DIA versus gateway exit, so it cannot detect SaaS-specific degradation beyond the overlay. Cloud OnRamp for IaaS with gateway discovery fails by concept because it is designed for connectivity to hosted compute in public cloud regions rather than SaaS front doors, using different discovery logic that does not track Microsoft 365 endpoints. As an exam caveat, distinguish SaaS exit optimization driven by cloud probes from transport optimization driven by overlay probes, since both are dynamic but answer different path questions. As a practical check, verify in SD-WAN Manager that the SaaS application is enabled with DIA and gateway sites defined, review probe loss and latency with the selected exit per site, then test failover by degrading branch DIA.