Final comprehensive exam

DP applies when a problem has overlapping subproblems AND…

• a sorted input
• optimal substructure
• no recursion
• a single base case

Which pagination style is most stable when rows are being inserted between page requests?

• Offset/limit
• Cursor/keyset
• Random sampling
• Client-side only

A mutex differs from a counting semaphore in that:

• A mutex allows N holders
• A mutex allows exactly one holder (mutual exclusion)
• A semaphore can't block
• They're the same

During a network partition, a strongly-consistent (CP) store will:

• Serve possibly-stale data to stay available
• Refuse some requests to avoid returning inconsistent data
• Always stay fully available and consistent
• Shard automatically

Write-back (write-behind) caching's main risk is:

• Slow reads
• Data loss if the cache dies before flushing
• It can't scale
• It always returns stale data

table_a has 3 rows; for a given a, b has 0, 1, 2 matches respectively. How many rows does `a LEFT JOIN b` produce?

• 3
• 0+1+2 = 3, plus the unmatched a → 4
• 6
• 9

Which lets you rank rows WITHOUT collapsing them into groups?

• GROUP BY
• A window function (e.g. RANK() OVER ...)
• DISTINCT
• HAVING

Given a composite index on (country, city, age), which queries can use it efficiently (leftmost-prefix rule)?

• WHERE country = ?
• WHERE city = ?
• WHERE country = ? AND city = ?
• WHERE city = ? AND age = ?

An ORM lazily loads each parent's children, firing 1 + N queries. This is the:

• Cartesian explosion
• N+1 query problem
• deadlock
• phantom read

T1 reads a row; T2 updates+commits it; T1 re-reads and sees a different value. This anomaly is a:

• Dirty read
• Non-repeatable read
• Phantom read
• Lost update

Which isolation level prevents dirty reads but still allows non-repeatable reads?

• READ UNCOMMITTED
• READ COMMITTED
• REPEATABLE READ
• SERIALIZABLE

Sharding primarily scales ___, while replication primarily scales ___.

• reads / writes
• writes & storage / reads & availability
• nothing / everything
• latency / nothing

INSIDE an fs.readFile callback, which fires first?

• setTimeout(fn, 0)
• setImmediate(fn)
• They're random
• Neither runs

What is logged?

async function f(){ throw new Error('boom'); }
try { f(); } catch (e) { console.log('caught'); }
console.log('after');

• caught
• after
• after, then an unhandled rejection
• boom

CPU-heavy image processing that must not block the HTTP server, sharing memory with the main process, calls for:

• a child process
• a worker thread
• cluster
• setTimeout chunks

Consistent hashing is primarily used to:

• encrypt data
• minimize key remapping when nodes are added/removed
• compress payloads
• enforce ACID

A Kubernetes Secret encrypts its data by default.

• True
• False

Hashing vs encryption:

• Both are reversible
• Hashing is one-way; encryption is reversible with a key
• Encryption is one-way; hashing is reversible
• Neither is reversible

An O(n²) algorithm takes 1s on 1,000 items. Roughly how long on 10,000 items?

• ~10s
• ~100s
• ~1s
• ~1,000s

BFS over a graph naturally uses which structure to track the frontier?

• Stack
• Queue
• Priority queue
• Hash set only

You need to keep keys in sorted order AND do range queries. A hash map is wrong because…

• it's too slow for lookups
• it has no inherent ordering
• it can't store many keys
• it uses too much memory

Exceeding a hash table's load factor triggers a resize/rehash.

• True
• False

Why do relational database indexes use B-trees / B+ trees?

• They use the least memory
• They're shallow (high fan-out) so few disk reads, and support range scans
• They're easier to code
• They guarantee O(1) lookups

You need the top-K largest items from a stream. Best structure?

• Sorted array
• Hash set
• Heap / priority queue
• Trie