AWS re:Invent 2018: Amazon DynamoDB Under the Hood: How We Built a Hyper-Scale Database (DAT321)
Why Hacker News recommends it
HN points to it as an authoritative look at DynamoDB internals and as evidence it is a complex distributed system.
What it is
An AWS re:Invent 2018 session on how DynamoDB was built as a hyper-scale NoSQL database. It discloses underpinnings of the service and how it is run for over 100,000 customers.
Recommended by 2 HN commenters across 2 different months, 2018 to 2022.
What Hacker News said
Recommends appwiz · 2019-06 · on “Learning to Build Distributed Systems”
Here are a couple videos from reInvent 2018: Jaso talking about DynamoDB internals [this video] Marc talking about Lambda internals [LINK]
Recommends mjb · 2022-01 · on “DynamoDB 10 years later”
We don't have a paper on DynamoDB's internals (yet?), but here's a talk you might find interesting from one of the folks who built and ran DDB for a long time: [this video] And Doug Terry talking through the details of how DynamoDB's transaction protocol works: [LINK] If we did publish more about the internals of DDB, what would you be looking to learn? Architecture? Operational experience? Developer experience? There's a lot of material we could share, and it's useful to hear where people would like us to focus.
Cites manigandham · 2018-12 · on “Amazon's cloud business is competing with its own customers”
DynamoDB is different from their published paper, which is mostly about designing a highly available key/value system that can be run on top of any other datastore. The paper does mention MySQL as a storage option. The recent MySQL info comes from this 2016 thread on DynamoDB storing empty strings: [LINK] Confirmed by this comment specifically: [LINK] The latest 2018 ReInvent deep dive on DynamoDB doesn't reveal anything but still fits if mysql is powering the storage nodes: [this video]
Cites talawahtech · 2019-01 · on “Amazon DocumentDB, with MongoDB compatibility”
Saying DynamoDB is built on top of InnoDB is a pretty big oversimplification of a much more complex distributed system[1] and for all we know they could have switched out the low level the storage engine on the backend to something like RocksDB or WiredTiger. The Aurora storage subsystem is much more limited in terms of horizontal scalability and performance, they probably chose it because it was a better/quicker fit. 1. [this video]