Pre-computed writes. Simple reads. 1M+ RPM at Kakao.
Currently on HBase, SlateDB planned.
→ github.com/kakao/actionbase
Added Actionbase as a read layer for friend relationships — without touching existing MySQL + HBase
· CDC via Debezium + Kafka
· Single schema for all query types
· Doesn't need to own data to add value
actionbase.io/stories/kakaotalk-friends/
Added Actionbase as a read layer for friend relationships — without touching existing MySQL + HBase
· CDC via Debezium + Kafka
· Single schema for all query types
· Doesn't need to own data to add value
actionbase.io/stories/kakaotalk-friends/
github.com/slatedb/slat...
And a WAL-based CDC PR as well:
github.com/slatedb/slat...
Community has several metadata RFCs up as well. 🚀 Some fairly big companies are evaluating it now, too!
github.com/slatedb/slat...
And a WAL-based CDC PR as well:
github.com/slatedb/slat...
Community has several metadata RFCs up as well. 🚀 Some fairly big companies are evaluating it now, too!
Sync? Latency spikes, backpressure, degradation.
Instead: queue to WAL, return immediately, process async.
→ actionbase.io/stories/kakaotalk-gift-recent-views/
Sync? Latency spikes, backpressure, degradation.
Instead: queue to WAL, return immediately, process async.
→ actionbase.io/stories/kakaotalk-gift-recent-views/
MySQL + Spring Batch + Redis → just Actionbase.
5 stages over months. No downtime, no data loss.
WAL replay + keeping rollback paths open.
→ actionbase.io/stories/kakaotalk-gift-wish/
MySQL + Spring Batch + Redis → just Actionbase.
5 stages over months. No downtime, no data loss.
WAL replay + keeping rollback paths open.
→ actionbase.io/stories/kakaotalk-gift-wish/
Embedded storage engine built on object storage
Embedded storage engine built on object storage
Pre-computed writes. Simple reads. 1M+ RPM at Kakao.
Currently on HBase, SlateDB planned.
→ github.com/kakao/actionbase
Pre-computed writes. Simple reads. 1M+ RPM at Kakao.
Currently on HBase, SlateDB planned.
→ github.com/kakao/actionbase