TIL about Apache Ozone.
Been thinking about a generalized implementation of Raft to pair with S3 similar to Neon’s Paxos layer. Ozone is the closest I’ve seen. This without the Hadoop cruft would enable more infra to run directly on S3.
/ht @_a__w_
https://t.co/jci13m5nzA
Andrej Karpathy is a legendary researcher who helped start OpenAI and created Stanford's first deep learning class.
@karpathy's advice on how to learn AI:
(1) 10,000 hours of deliberate practice will make you an expert. You can iterate as you work. Only compare yourself to the past, not to others.
(2) Don't worry about making mistakes. You build intuitions on what is useful vs. not useful- they are not dead work.
(3) Teach to strengthen your understanding and find gaps in knowledge. Similar to "If you can't explain it to a six-year-old, then you don't understand it yourself" - Albert Einstein.
@jaykreps Won't this put Kafka in the critical request path for all business services writing to it? Also, even high latency across the 2PC can cause issues for business services. That's probably one advantage for using CDC to capture your data through a business database.
How to become expert at thing:
1 iteratively take on concrete projects and accomplish them depth wise, learning “on demand” (ie don’t learn bottom up breadth wise)
2 teach/summarize everything you learn in your own words
3 only compare yourself to younger you, never to others
Any presentations or blog posts that do a deep dive comparison between Apache Flink and Kafka streams/ksqlDB? So far, here is what I've found:
https://t.co/lXhwygSJHW
"This commit allows to optimize performance by
providing a new optional mode of TLS sendfile(), in which the extra copy
is skipped. Removing this copy improves performance significantly, as
TLS and TCP sendfile perform the same operations"
https://t.co/VF2f2O43M7
@JSchabowsky Thanks, yes, that does make sense. I guess once you go down the route of offering Kafka as a service you'll need to enforce quotas to manage capacity between the different tenants. That can get mess though if a critical service needs more capacity and it's not available.
If you build streaming services based on Change Data Capture (CDC), how do you ensure an upstream DB migration doesn't break the downstream consumer logic?
Seems like domain events create a more decoupled architecture, what am I missing?
https://t.co/6NxG2grrD6
@gwenshap@__Tommm Hey @gwenshap, what are your thoughts on an L3/L4 proxy like envoy? This would simplify things on the client as you could continue to use the native Kafka client libraries.
@gwenshap As far as I could tell (last time I checked) quotas were still not supported in the Admin/Client API, any plans on adding that functionality?