Ah @renerocksai wow, just saw this and itās incredible.
What a journey. I loved reading the writeup, all the plot twists and turns, and how you systematically swapped components to arrive at the architecture.
Thank you so so much (and elaborating on my tweet!). š
Your Sunday Read
Read how @renerocksai built a high-throughput ticket sales system on @TigerBeetleDB, covering domain modeling, technical decisions, and throughput optimization.
The Sunday Surprise
When adding CPUs slows you down!
Every successful career I've ever known was filled with long periods of meandering, months or even years when no one knew what would happen next.
Look at me: I started as a geology major turned failed realtor.
I then ended up founding several companies, one of which ended up becoming @netflix
Life is inherently uncertain. I think making friends with that fact can really help us.
Try things, see what happens, see what you like and dislike, try again.
This is the way.
By utilizing a low-level engineering approach, @DorLaor and @AviKivity believed they could squeeze considerably more power from the underlying infrastructure. Felipe Mendes looks at the capabilities that set us apart from Cassandra. ow.ly/OuxO50X2ONu#ScyllaDB
āIn August 2024, tickets for the Oasis reunion tour sold out rapidly, with approximately 1.4 million tickets sold over six hours [ā¦] or 65 tickets per second for a high-demand ticket sale scenario.ā
A phenomenal post by @renerocksai on how he 10xād this with @TigerBeetleDB.
š Your Weekend Read: Whether you're looking to run your first queries or running it in a production environment, ScyllaDB in Action is a practical guide for building apps with #ScyllaDB. Learn everything about our database from @Boluptuous. ow.ly/Gl0u50XopsY
Here is the slide deck I used for my sessions at @MongoDB .local Paris, covering topics such as transactions, indexes, common myths, and a comparison with PostgreSQL:
mdb.link/paris-slides-franck
You can find the recorded demos here:
mdb.link/paris-demos-franck
Bātrees face issues like page splits, locks, and buffer pins. LSMātrees avoid them but raise write amplification. WiredTiger takes a third path: lockāfree Bātrees with hazard pointers for reads and skip lists for writes. Great article on the @MongoDB blog:
mongodb.com/company/blog/engā¦
The most complex part of almost any application is state management, especially between the frontend app and the backend database. The primitive you want is sync.
Also probably you want this swag.
We had some unique goals when we built Vector Search for #ScyllaDB. We noticed that most built-in solutions sacrifice performance for accuracy or scale, so we decided to design our architecture to deliver all three. Find out how here > ow.ly/Yq3y50XjMqM
imgui-react-runtime is going public on GitHub on Friday morning PST. It is not fully ready, but I guess it will never be - there is always one more thing to cleanup.