BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//sf26-agenda//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
BEGIN:VEVENT
UID:SESS-82@sf26-agenda
DTSTAMP:20261020T151709Z
DTSTART:20261105T183000Z
DTEND:20261105T191500Z
SUMMARY:EventBus at Dropbox: Self-Serve Millions of Events Per Second with 
 Near-Zero Ops
LOCATION:Room 2010
DESCRIPTION:David Yun\n\nWhen operating 6 platforms with a team of 3 and ac
 tively developing the 7th\, near-zero ops is no longer a luxury: it is a n
 ecessity. In this talk\, we deep dive into Dropbox's standard event proces
 sing platform: EventBus. Today EventBus processes hundreds of billions of 
 events per day across 50+ use cases\, serving as the backbone for data pip
 elines\, search indexing\, and AI ingestion company-wide. The platform abs
 tracts away the complexities of operating Kafka consumers and transforms K
 afka consumption into familiar stateless remote procedure calls focusing o
 n business logic. Users of the platform are able to self-onboard use cases
  that consume millions of messages per second\, and everyone still gets to
  sleep at night. EventBus carries the golden standard in platform reliabil
 ity at Dropbox\, with 0 SEVs incurred since GA >1 year ago despite having 
 0 resources dedicated for KTLO. This talk covers advanced topics such as g
 oing beyond partition-level parallelism while retaining message key orderi
 ng\, how we deal with hot partitions\, protecting tenants from noisy neigh
 bor issues\, and how we are able to operate cost effectively on top of bei
 ng operationally light. Whether you already run Kafka at scale or are weig
 hing whether to build a platform layer over it\, you will leave with a con
 crete blueprint for turning per-team consumer sprawl into a paved road eng
 ineers actually want to use.
URL:https://sf26.rmoff.net/s/eventbus-at-dropbox-self-serve-millions-of-eve
 nts-per-second-82
END:VEVENT
END:VCALENDAR