How Netflix Uses Kinesis Streams to Monitor Applications and Analyze Billions of Traffic Flows

[@chrispaul7890]

At about 25:00 he you start talking about the issues with using cache and cache invalidation. You mention that if you set the TTL too long you will have stale (unreliable) data. Aren't you just pulling in events effectively as historical transactions? How does this get invalidated with time? What am I missing?

[@muthukumarkusalavan3261]

i also had the same doubt.

[@MrMylonz]

@@muthukumarkusalavan3261 I think he is indicating that during the enrichment process of resolving the historical transaction's IP address into something more meaningful (which may change over time) by a lookup to a cached value it may be a stale value if that cache not updated quickly enough. Hence a transaction gets attributed to the wrong resource.

[@AbHIShEkUPAdhYaYshekup]

The talk is good but I didn't get towards end if Netflix is using Kinesis or Kafka

[@solderbuff]

29:08 - Both.

[@belisario93]

not nearly enough people laughed at 28:48

[@TouchedAlot]

don't cross the streams!

[@xmsmdmx]

Apparently, nothing strange ever happened in their neighborhoods.

[@Eskimoz]

Nice.

[@staj]

The Netflix Use Case bit is Brilliant! Very insightful Video! Thanks.