Awesome Talks

List of my personal favourite talks on Youtube

All time favs#

Title Notes Topic
[Facebook] USENIX ATC 2013 - TAO Facebook share how they build their own graph database, supporting billions of QPS in 2010s Database, Caching
[New Relic] The Evolution of a Planetary-scale Distributed Database New Relic share how their architecture evolved from the beginning to become Architecture, Evolution
[Shopify] 5 Years of Rails Scaling to 80k RPS Shopify explains how their system evolved from a simple monolithic application to become a sharded (cell-like) system Architecture, Evolution, Ruby on Rails
[Facebook] RocksDB A revolutionary DB was born. Many of the databases built today (2024) are actually based on this RocksDB Database
[Facebook] Mcrouter Facebook shares how they built a memcached proxy router that help them scale to billions of QPS. The layered topology is what makes mcrouter awesome! Database, Caching
[Microsoft] Beyond FaaS: Stateful, Scalable, & Scrutable Cloud Apps, [Microsoft] Using Project Orleans to Build & Scale Halo Generally, everyone try to avoids writing & managing stateful workloads because they are difficult to handle. But with Orleans, you get so much benefits from writing stateful application. This is first used in Halo and now powers many of Azure cloud services. Architecture, Actor Model
[Slack] KalDB: A cloud native log search platform Slack explains the problem they face with huge amount of logs and how they architect the new system to solve that Database
[AWS] Amazon S3 Deep Dive AWS explains how building object store is complicated Scalability, Storage
[AWS] USENIX ATC 2022 - Amazon DynamoDB The architecture of DynamoDB Database
[AWS] Handling Log4Shell Fascinating to know how they (especially Principal engineers) operates at AWS and how they react to critical security flaws. Security, Process
[AWS] Cell-based architecture AWS explains how they architect their system to be resilient to failures Architecture
[Microsoft] Outlook Service Microsoft explains how they build Outlook mail in 2015 that scales. Their system looks like cell-based architecture. Architecture
[Datadog] Building highly reliable data pipelines Datadog explains how they make a tradeoff calls to build their data pipeline. IMO, they’re taking a different approach by building many ephemeral clusters instead of baby-sitting several long-lived clusters. Architecture

* Many of the AWS Reinvent talks presented by their Distinguished & Principal engineers are awesome!

Pretty good ones#

Title Notes Topic
[Facebook] ZippyDB Facebook explains how they’re building a strongly consistent KV store using RocksDB. They also talks about their replication/log system. Database
[Pinterest] Rocksplicator Pinterest are doing interesting stuffs with RocksDB Database
[Facebook] Cache Made Consistent Facebook explains how caching at their scale is not easy Caching
[Soundcloud] Consistency without consensus in production systems Using the first principles, Soundcloud is building their Feed differently compared to other social networks, using CRDT. Architecture, CRDT
[Martin Kleppman] CRDT A pretty cool tech to build collaborative tool Database
[Sasa Juric]The Soul of Erlang and Elixir See how Elixir can handle concurrency pretty well Elixir