Quick Revision

GK One-Line Question & Answer

15541+ short questions with short answers, covering every category and sub-category on the site — no long articles to scroll through. Good for a fast recap before an exam, or a few minutes of daily practice.

Looking for exam process/strategy questions instead? Try the Exam Preparation FAQ →

DBMS → Introduction to DBMS 30

Exclusive (X) lock allows
ONE transaction to read or write blocking all others
click to copy
Write-Ahead Logging (WAL) requires
Log written to stable storage BEFORE data page written to disk
click to copy
Redo in recovery
Re-applies operations of committed transactions to ensure durability after crash
click to copy
Undo in recovery
Reverses operations of uncommitted transactions to restore consistency
click to copy
ARIES recovery passes in order
Analysis, Redo, Undo
click to copy
Checkpoint in recovery is used to
Reduce redo work needed after crash
click to copy
No-Force policy means
Pages need NOT be forced to disk at commit - requires REDO on recovery
click to copy
Steal policy means
Dirty pages of uncommitted transactions CAN be written to disk - requires UNDO on recovery
click to copy
Optimistic CC performs best when
Conflicts are RARE - mostly reads few conflicts
click to copy
Multiple granularity locking allows
Different hierarchy levels: database, table, page, row
click to copy
IS (Intention-Shared) lock signals
Intent to set S locks at FINER granularity levels
click to copy
IX (Intention-Exclusive) lock signals
Intent to set X locks at FINER granularity levels
click to copy
Recoverable schedule ensures
T_j commits only AFTER all transactions whose data it read have committed
click to copy
Cascadeless schedule prevents
Cascading rollbacks by allowing reads ONLY of committed data
click to copy
ARIES Redo pass
Repeats ALL logged operations to bring DB to exact crash-time state
click to copy
ARIES Undo pass
Reverses loser (uncommitted at crash) transaction operations in reverse log order
click to copy
Log Sequence Number (LSN)
Uniquely identifies and orders each log record (monotonically increasing)
click to copy
Fuzzy checkpoint allows
Active transactions to continue while checkpoint record written to log
click to copy
The 'lost update' problem occurs when
Two transactions update same item and one update is overwritten by the other
click to copy
ARIES uses which buffer management policy combination
No-Force + Steal
click to copy
Rigorous 2PL holds
ALL locks (S and X) until transaction commits or aborts
click to copy
A database block/page is
Unit of data transfer between disk and memory (fixed-size)
click to copy
Seek time is
Time to move disk arm to the correct track
click to copy
Rotational latency is
Time for correct sector to rotate under read/write head
click to copy
Buffer pool is
Portion of main memory caching database pages
click to copy
LRU buffer replacement replaces
LEAST recently used page (oldest access time)
click to copy
Query cost is primarily measured by
Number of disk I/O operations
click to copy
Query optimization selects
Most efficient (lowest cost) execution plan from equivalent alternatives
click to copy
Pushing selection (σ) down operator tree
Reduces intermediate result sizes - fewer tuples flow to expensive operators
click to copy
Pipelining in query processing
Passes tuples directly from one operator to next without full materialization
click to copy

DBMS → Transactions 3

Cascading rollback occurs when
Rolling back T1 forces other transactions that read T1's dirty data to also roll back
click to copy
Timestamp-based CC assigns each transaction
Unique timestamp at start time for ordering operations
click to copy
Transaction state 'aborted' means
Rolled back and database restored to pre-transaction state
click to copy

DBMS → Indexes 5

A clustered index stores
Data in same sorted order as the index
click to copy
A dense index has
One index entry for EVERY record in the data file
click to copy
A sparse index has
Index entries for ONLY SOME records (typically one per data block)
click to copy
B+ tree supports
Both equality AND range queries with O(log n) performance
click to copy
Hash index is best for
Equality queries (exact match =)
click to copy

DBMS → Joins 2

Sort-Merge Join requires
Both relations sorted on join attribute
click to copy
Hash Join build phase
Hashes SMALLER relation into memory hash table
click to copy