AWS re:Invent 2017: DynamoDB adaptive capacity: smooth performance for chaotic workl (DAT327)

[@rickfarina9835]

Super, super presentation and Q and A. Definitely worth taking the time to watch and understand. Thanks very much Rick.

[@sheelstera]

At 25:00 , when he says that 450K units of unused RCUs have piled up in last 5 mins in the burst bucket and are available for use @3K RCU, then how can 450K units be drained @peak rate of 3K RCU in 5 minutes; 450K units being disbursed @3K RCU would be 150 seconds, right? Similiarly, other calculations are also not matching. Am I understanding this wrong?

[@swayampb]

if you refer the dynamoDB doc, it says 1 RCU is basically 1 strong consistent read per second or 2 weak consistent read per sec. So i think for the presentation they have taken the liberty to chose the weak reads. you can check all that in the doc : docs.aws.amazon.com/amazondynamodb/latest/developerguide/HowItWorks.ReadWriteCapacityMode.html#HowItWorks.requests

[@travisbaker]

"spurious throttling" would be a good name for a rock band.

[@MarkRose1337]

🤘

[@rodeffect]

Are there any further docs on Adaptive Capacity? This is a really great new feature that solves some major pain points of our DynamoDB usage, it's surprising that it was launched with so little fanfare!

[@sidhanthweek]

docs.aws.amazon.com/amazondynamodb/latest/developerguide/GuidelinesForTables.html#GuidelinesForTables.AdaptiveCapacity

[@sidhanthweek]

Documentation: docs.aws.amazon.com/amazondynamodb/latest/developerguide/GuidelinesForTables.html#GuidelinesForTables.AdaptiveCapacity