medriscoll
Re:What Loggly Learned About Scaling with Apache Storm
on Thursday April 03, 2014 @08:31PM (#7528267)
Nice piece and love the "no log left behind" philosophy.
Recognizing every company's use cases are different, we've found Storm has worked quite well for us. We run an always always-on ingestion service with sustained peaks north of 1 million EPS.
We presented a talk at Strata two months ago on the Kafka --> Storm --> Druid architecture that we rely on: http://youtu.be/kJMYVpnW_AQ .
Re:Redis new data structure: the HyperLogLog
on Tuesday April 01, 2014 @05:55AM (#7507126)
The Druid.io folks recently blogged about three additional optimizations to HyperLogLog that compact the registers and can improve query speed:
https://groups.google.com/forum/#!searchin/druid-development...