BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//sf26-agenda//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
BEGIN:VEVENT
UID:SESS-181@sf26-agenda
DTSTAMP:20261020T151709Z
DTSTART:20261105T193000Z
DTEND:20261105T201500Z
SUMMARY:Driving the Enterprise Data Highway: Why McDonald's Ditched Self Ma
 naged Kafka
LOCATION:Room 2006
DESCRIPTION:Yinsong Li\, Raj Bhino Rajamani\n\nAs enterprises scale globall
 y\, self-managed streaming platforms often struggle with operational compl
 exity\, cost\, and real-time data demands. In this session\, we share how 
 McDonald’s modernized its data streaming platform by migrating over 200+
  workloads and 2\,000+ topics across multiple kafka clusters processing ov
 er 300 million events/day (~25 TB/day) to a fully managed\, cloud-native s
 treaming service as part of a broader GCP transformation. Beyond the migra
 tion effort\, we established a centralized Enterprise Data Highway as our 
 foundation; designed to connect 40\,000+ edge locations\, digital channels
 \, and IoT devices enabling real-time\, event-driven data streaming across
  the enterprise where the data is always served fresh and fast while layin
 g the groundwork for McDonald’s NEXT and proving the future of innovatio
 n is something we’re all lovin’. This session will highlight key archi
 tectural patterns\, including bi-directional data replication\, hybrid pub
 lic/private cluster strategies\, large-scale schema governance\, future pr
 oofing\, and robust edge-to-cloud synchronization with built-in offline re
 silience. Relevance to Real-World Data Streaming: Many enterprises face si
 milar challenges: fragmented kafka deployments\, high operational overhead
 \, limited scalability\, and difficulty enabling real-time use cases acros
 s distributed environments. This session provides insights on how to tackl
 e these challenges: ** How to transition from self-managed to fully manage
 d Kafka without disrupting business operations at scale ** How to design s
 ecure\, multi-region streaming architectures with hybrid connectivity ** H
 ow to support edge-to-cloud patterns at massive global scale with unstable
  network conditions using NATS connectivity to Confluent Cloud ** How to m
 ove from the Wild Wild West to an established governed\, reliable data str
 eaming platforms that enable enterprise-wide reuse and making it future re
 ady for Agentic consumption and predictive Inferences Audience Takeaways: 
 This session provides actionable insights for architects and platform team
 s looking to reduce operational overhead\, execute complex and incremental
  migration with no impact to business continuity. Also focuses on concrete
  architectural and operational patterns including: ** Enabling an Enterpri
 se Data Highway\, migrating existing topics from MSK to Confluent Cloud us
 ing replicators\, cluster linking and on-demand bi-directional sync ** Clu
 ster isolation strategies (public/private) to enforce security\, tenancy\,
  and data segmentation ** Schema governance models for enforcing compatibi
 lity and evolution across domains at scale\, cost optimization using compr
 ession to move data efficiently ** Edge-to-cloud synchronization patterns\
 , including buffering and recovery mechanisms for intermittent connectivit
 y ** Incremental migration approach with on-demand bi-directional sync\, z
 ero-downtime cutover strategy for idempotent and non-idempotent consumers.
URL:https://sf26.rmoff.net/s/driving-the-enterprise-data-highway-why-mcdona
 lds-ditched-181
END:VEVENT
END:VCALENDAR