Vivcre Learn
- System Design
CAP: What a Partition Actually Forces You to Choose
Why, when the network splits, a replica has exactly two options — and why the real choice is CP vs AP, not 'pick two of three'.
- Philosophy
Ep 5: Jain Philosophy — जैन दर्शन
Reconstruction of Jain philosophy — anekāntavāda, syādvāda, the soul Chārvāka denied, and karma as matter — as a draft to verify against Divyakirti's lecture.
- Database
Three Join Algorithms, and When Each Wins
Nested-loop, hash, and merge joins exist because the cheapest way to combine two tables depends on their sizes, indexes, and sort order.
- Kafka
Kafka Consumer Groups: Rebalancing, and Duplicate vs Skip
How a group detects a dead (or just slow) consumer, who inherits its partitions, and why committing before or after processing decides whether you lose messages or process them twice.
- Kafka
Kafka Exactly-Once: At-Least-Once Plus Dedup
Why 'exactly-once' can't mean delivered once, and how Kafka's idempotent producer and transactions make duplicates invisible — inside Kafka.
- Kafka
Kafka Is a Log, Not a Queue
A Kafka partition is an append-only log; consumers don't remove messages, they advance their own offset — which is why many can read the same stream and replay it.
- Kafka
Kafka Partitions and Replication: Where Availability Is Spent
Splitting a topic into partitions buys parallelism at the cost of global ordering; replication plus acks decides durability versus availability.
- System Design
PACELC: The Tradeoff That's Always On
CAP only describes the rare partition. PACELC adds the truth that even on a healthy network, every read trades latency against consistency.
- Philosophy
Ep 6: Buddha's Life Story (Buddhism Part 1) — बौद्ध दर्शन भाग-1
Reconstruction of the Buddha's life as turning points, each raising the question that his philosophy (Part 2) will answer. Narrative, not argument. Draft to verify against the lecture.
- Philosophy
Ep 7: Buddha's Philosophy (Buddhism Part 2) — बौद्ध दर्शन भाग-2
Reconstruction of Buddhist philosophy — the Four Noble Truths, impermanence, no-self (anātman), dependent origination, and nirvāṇa. Draft to verify against the lecture.
- Philosophy
Ep 2: Indian Philosophy — A Map (Vedas to Osho) — भारतीय दर्शन: एक परिचय
Reconstruction of the survey lecture — how Indian philosophy is classified into āstika and nāstika (and why that is NOT theist vs atheist), the six + three schools, and the threads they share. Draft to verify against the lecture.
- Philosophy
Ep 3: The Search for Ultimate Reality — अंतिम सत्य क्या है
Reconstruction of the metaphysics lecture — the One-vs-Many and permanence-vs-change problem, appearance vs reality, and how the schools answer 'what is ultimately real'. Draft to verify against the lecture.
- Philosophy
Ep 1: What is Philosophy? — दर्शन क्या है
Reconstruction of the framing lecture — what philosophy is, why the Indian word darśana means 'a way of seeing', and how it differs from science and religion. Draft to verify against the lecture.
- Database
Why an Index Turns 200,000 Reads Into 4
How a database physically finds a row, why un-indexed lookups scale linearly, and how a B-tree collapses that to a handful of page reads.
- Database
Composite Index Column Order: Equality First, Range Last
Why the column order in a multi-column index decides whether the planner can use it, and the equality-first, range-last rule that follows from how the index is sorted.
- Database
Covering Indexes and the Index-Only Scan
How an index that contains every column a query needs answers it without ever touching the heap — and the MVCC caveat that quietly breaks the trick.
- Database
Compaction: The Triangle You Can't Escape
Size-tiered vs leveled compaction, and why every LSM engine must pay one of read, write, or space amplification.
- Database
MongoDB Data Modeling: Start From the Workload
Why MongoDB inverts relational modeling — shape the data around the queries — and how embed-vs-reference is the same read/write tradeoff you already know, one level up.
- Database
From Lost Updates to MVCC: Isolating Concurrent Transactions
How a naive concurrent counter loses updates, why pure locking creates a traffic jam, and how multi-version concurrency control lets readers and writers stop blocking each other.
- Database
How LSM Trees Make Writes Cheap
Why write-heavy databases buffer in memory and merge sorted files on disk.