I've heard folks cite this as a good resource - www.cs.cmu.edu/~harchol/Per...
03.01.2025 19:26
👍 1
🔁 0
💬 0
📌 0
I've heard folks cite this as a good resource - www.cs.cmu.edu/~harchol/Per...
Just picked this up today! Looks like it's going to be a good read.
I want to read a few papers over the holidays. What were the most interesting distributed systems or database papers you read this year? 👀
"Last Xmas
I rsynced you my heart
And the very next day
rm -rf ~/heart
This year, to save me from tears
sudo chmod -r 000 ~/heart"
Moderating this wrap up session was such a fun and insightful experience - thank you @hasgeek.com for the opportunity 😁
This in-memory implementation currently relies on a "collapse" function to remove older data; it would be nice to have this functionality as part of the iterator itself.
A rust code snippet showing a test that demonstrates the process of compaction in an LSM Tree
Finally got Merge Iterators working 🫡
LSM Tree Compaction largely relies on merge iterators, which take in multiple SSTables and give you a new SSTable which doesn't have any old copies of data present. See how we went from 7 to 3 records - both are the exact same state!
Immutability has the added benefit of simplifying WALs; in an LSM Tree, your logical log and your physical log are the same!
see: Database Internals, page 91
Computer screen with a pomodoro timer on it with code in the background
It's just me and my pomodoro timer against the whole world
Super pumped to see how this turns out 😁🤘🏼
There are 2.4999999999994 hard things in computer science: naming, floating point math and exactly once delivery.
This is another really awesome project! Watching @jonhoo.eu's talk about this was what first fascinated me with the space. I highly recommend their deep dive videos on this, really worth a watch - thank you for sharing 😁
This is an awesome watch - www.youtube.com/watch?v=eYJA...
Incremental view maintenance is such an interesting problem. The composition proof in the video blew my mind! This is some really awesome work by all the authors and contributors @feldera.bsky.social.
Just graduated yesterday!
I've been tinkering, breaking, and fixing things since I was a child, but today, I'm officially an Engineer - with a Bachelor of Technology in Computer Science and Engineering.
It's a strange, new feeling, and a strange, new world - I'm excited to see where this goes!
😭
One of my favorite things to do is to start typing out the lyrics to a song on a WhatsApp group. When someone joins in, it's like we're singing that song together :)
Hi Deepak! We are both on discord and telegram. You can find all the links on our Hasgeek page -
hasgeek.com/bengalurusys...
Today, we just wrapped up an informative+awesome talk by @alexgarcia.xyz on sqlite-vec at the Bengaluru Systems Meetup.
It covered SQLite extensions, vector search, shadow tables (very cool!), hybrid search with rank fusion, binary/scalar quantization
Recording (Ty Alex!) -
youtu.be/GpTOsTxuLLA
👋🏼
Don't miss this! Alex will be recording this too 😁
So far:
☑️ working point reads + writes (in-mem + disk)
☑️ JSON SSTables
☑️ BTreeMap memtables
☑️ working flush path (w/memtable swap)
☑️ Single-level Tree
Can't wait to implement the compaction codepath ⚡all the code is in a single file so far.
Crashing on invariant violation has helped me catch a few interesting bugs that I didn't know existed
Make art that you think is fuckin sick as hell. Make the shit you never thought you could that you always wanted to see. Beat your insecurity to death and fight to become your own favorite artist.
This would help SlateDB writes, right? Or do you feel like the gains from batching would be lost by using these appends?
Hi! Navin.
This is so cool!
I hear you - working on it!
I'd love to be on this list too - new to bsky, and I work on the Magma LSM-based Storage Engine at Couchbase - thanks a ton, Chris 😁
A week later and no progress; I've been losing a lot more time than I want to admit to just doomscrolling. That changes today ⚡
Go Phil! ⚡