Paxos

in many ways, my career arc is kind of similar to yours where I was more a distributed systems geek, having built lots of large-scale systems.

And then I saw at some point But, hey, having built these large-scale systems and Paxos that works across multiple data centers, having built all that for DynamoDB to S3 to everything else, let alone DynamoDB and several others, I felt like, okay, hey, now, yes, you can continue to actually make it scale to the infinite scale that you keep scaling.

But the next big set of challenges were related to AI and how unexplored that dimension was.

Personally, having gone through all this journey from data to analytics to AI, now with agentic AI, the single biggest thing I always tell practically everybody is the only thing you can do to slow down your career is you stop learning.

To me, computer science also involves like understanding things and also operating system compilers and actually all things related to even database internals and all these things.

Yes, code generation can be done, but reasoning through various aspects of system and actually being able to operate at the next layer, even when AI agents generate code and test it, for you to be able to reason at the next layer is still important.

So it's a little bit like, yes, distributed systems, not many people need to write Paxos code anymore, but people still need to reason through on, hey, and I'm training the post-training this large cluster and improving the packets per second on this training improvement is significant dollar savings.

knowing the fundamentals always, always matters, whether your passion is in computer science or physics or chemistry or any of it more than anything else.

Envartolome.com.

The article lists key papers that shaped the field of distributed systems, including foundational works by Leslie Lamport, The Byzantine General's Problem, Paxos, and Bitcoin's White Paper.

It aims to provide a starting point for understanding how distributed systems evolved and highlight influential research papers across decades.

In the comments, the community mostly discussed the importance of Lamport's work and his influence on distributed systems.

Users debated the readability of Lamport's papers, with some finding his writings less accessible, while others appreciated their clarity.

There was considerable discussion about the relevance of classical papers like Paxos and Byzantine fault tolerance, and whether newer approaches have replaced them.

Several comments highlighted that the core principles, including consensus and logical clocks, remain vital, and some shared personal experiences studying distributed systems.

Overall, the community viewed these papers as essential foundations, with some noting ongoing relevance in both theory and modern applications.