Omar

Why CPU utilization based Load balancing increase latency and errors? In modern application system, compute runs on VM with many other VM on a host machine. We call other VM that runs with our application antagonist. Now come to an example:

@its_ani9 @0xlelouch_ Hyperloglog gives approximate result

@0xlelouch_ For 10TB clickstream data, I’d use HyperLogLog instead of HashSet. It estimates distinct user IDs using hashed values with very low memory. Process data as a stream/distributed job (Spark/Hadoop). Time: O(N), memory: very small, scalable for billions of records.

Interested in Data structures? Count the number of distinct user IDs across a 10TB clickstream. Solve that!

@SumitM_X Omar says: it does not makes sense. It can be same user but always true.

Tech Lead speaks in the meeting: "If two requests come from same IP and port, they are the same user." Manager says : "Makes sense" What do you say?

How a simple #caching strategy if not careful can brought a black swan events or #Metastable failure for your service. I wrote a summary based on a paper(sigops.org/s/conferences/…). nanofaroque.github.io/personal_prese…

@nehanagarr Lots of romantic movie by SRK.

The USA has Technology. China has Manufacturing power. What does India have?

I often asked why we should learn in the age of AI. Here is my thought in this article, please read it and share your thoughts.

Do you remember when you joined X? I do! #MyXAnniversary Wow

@SumitM_X String is not the right choice here however It can be like: private static final Object lock = new Object();

Is this code ok ? String lock = "LOCK"; synchronized (lock) { // critical section }

@javarevisited Order is not guranteed here.

Java Stream Interview ☕ What happens here? Stream.of(1,2,3,4) .parallel() .forEach(System.out::print); Is output order guaranteed?

@arpit_bhayani The Non-Deterministic Bomb Ticking in Your Codebase. I shared my concern here: nanofaroque.github.io/personal_prese…

With agentic slop, we are trading software reliability for shipping velocity and calling it progress. It isn't. Systems are more fragile than ever, and engineers building them no longer trust their code to hold up in real-world edge cases. I am pro-AI, but this will backfire - big time.

@ashoKumar89 @javarevisited I thought WhatsApp does not even save the message outside of the device .

@javarevisited They do not load everything. - Only recent chats/messages are fetched first. - Data is paginated and lazy loaded as user scroll - Uses local storage/cache on device - Backend uses indexed queries + shrading So it feels instant because it loads just what you need first

As a developer, have you wondered: When you open WhatsApp, your entire chat history loads almost immediately. For billions of messages stored… How is this data fetched so quickly?

@EOEboh You need frequent read and stale data is acceptable. This is a general case.

Simple question for engineers: How do you decide when to add a cache vs when not to?

@javarevisited Also, they might have support from the solution architect to solve the problem. Lol

@javarevisited Because they offload everything on the cloud.

There are engineers who can design systems serving millions… but can’t get hired because they didn’t practice interview puzzles.

@LauraLoomer You are not a parent, are you?

Is there anything worse than a crying baby on a plane? I wish parents would control their children. It’s so disruptive. I refuse to believe a baby cries for 10 hours. At some point this is just bad parenting, right?

Nice explanation by Claude: Lock = a key to a room (only one key, only the holder can leave and hand it back) Semaphore = a bouncer with N wristbands (anyone leaving can hand their wristband to the next person)

@hiarun02 Cache keeps data on dram which is ephemeral, restart just wipe everything out normally unless you use other way to keep it safe. Database is persistent storage.

Backend Developer interviewer : If cache is faster than a database, why isn’t everything stored in cache ?

@SumitM_X People talked about perception, calculation. I want to add: Memory because 500 can be represented as a Boolean flag(>=500 or not), single byte instead of integer 4byte or more. That reduce rpc latency tax and memory requirements.

As a developer, Have you ever wondered: WHY in LinkedIn… you see "500+ connections" Why not exact number? What trade-off has LinkedIn made?