How Netflix uses Druid for realtime insights
netflixtechblog.com
How Netflix uses Druid for realtime insights
1–10 of 60 posts
Re: How Netflix uses Druid for realtime insights
#2Can anyone from Netflix comment on why not clickhouse? We tried druid, and the performance was pretty bad compared to a clickhouse.
Re: How Netflix uses Druid for realtime insights
#3> During software updates, we enable the new version for a subset of users and ... compare how the new version is performing vs the previous version. Any regression in the metrics gives us a signal to abort the update and revert
How do you account for the possibility that the update only performs badly because it’s different than what users are used to, but would actually be an improvement in the long run?
Re: How Netflix uses Druid for realtime insights
#4Is it possible to view this article without creating a medium account?
Re: How Netflix uses Druid for realtime insights
#5Is it possible to view this article without creating a medium account?
It looks like Medium is opening their paywall modal based on viewed articles (and cookies). Therefore, you could try private browsing.
Re: How Netflix uses Druid for realtime insights
#6> During software updates, we enable the new version for a subset of users and ... compare how the new version is performing vs the previous version. Any regression in the metrics gives us a signal to abort the update and revert How do you account for the possibility that the update only performs badly because it’s different than what users are used to, but would actually be an improvement in the long run?
Have you worked at a large company before? You don't!
Re: How Netflix uses Druid for realtime insights
#7> How can we be confident that updates are not harming our users?
Maybe this is why they finally figured out the auto-playing trailer stuff was a horrible change (I cancelled my account over that). One can only hope.
Re: How Netflix uses Druid for realtime insights
#8Can anyone from Netflix comment on why not clickhouse? We tried druid, and the performance was pretty bad compared to a clickhouse.
Can confirm clickhouse is generally faster across most typical workloads
Re: How Netflix uses Druid for realtime insights
#9> During software updates, we enable the new version for a subset of users and ... compare how the new version is performing vs the previous version. Any regression in the metrics gives us a signal to abort the update and revert How do you account for the possibility that the update only performs badly because it’s different than what users are used to, but would actually be an improvement in the long run?
You acknowledge that fact and if you're making a severe change you either run a longer test or you break up the large change into manageable chunks (Google Maps is a great example of this).
For something that is effectively binary and can't be changed incrementally the only way is to extend the evaluation window or make a call with limited information.