Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn from Google's real-world experience deploying network Quality of Service (QoS) alignment to application priorities for storage services in large-scale data center networks through this 16-minute conference presentation from NSDI '25. Discover how to properly configure network QoS settings in packet headers to ensure application network traffic receives appropriate prioritization within data center infrastructure. Explore the challenges and unexpected insights encountered during fleet-wide deployment, including the counterintuitive finding that downgrading traffic to lower QoS classes doesn't always result in worse network latency due to over-utilized QoS bands. Examine the technical approach of mapping application priorities to DSCP bits in IP headers on a per-RPC basis for intra-cluster storage traffic, and understand how network switches and routers interpret these configurations to allocate resources like buffer space and scheduling priorities. Gain practical guidance on implementing QoS-based scheduling strategies to meet service guarantees, addressing performance regression concerns, and scaling these solutions across networks of any size based on lessons learned from Google's production deployment experience.
