Technical Terms — Backend & System Design

A1

ACID

ACID (Atomicity, Consistency, Isolation, Durability) defines four properties that guarantee database transactions are processed reliably, even during crashes, errors, or concurrent access.

Read definition

B4

B-Tree

A B-tree is a self-balancing tree data structure that maintains sorted data and allows searches, insertions, and deletions in logarithmic time, forming the foundation of most database indexes.

Read definition

Backpressure

Backpressure is a flow control mechanism where a slow consumer signals upstream producers to reduce their sending rate, preventing unbounded queue growth and resource exhaustion.

Read definition

BASE

BASE is an acronym contrasting ACID for distributed systems: Basically Available, Soft state, Eventually consistent. It describes a model where availability is…

Read definition

Bloom Filter

A Bloom filter is a space-efficient probabilistic data structure that tests set membership, answering "definitely not in set" or "possibly in set" with tunable false positive rates but zero false negatives.

Read definition

C13

Caching

Caching stores frequently accessed data in a faster storage (often memory) to speed up retrieval.

Read definition

CAP Theorem

The CAP theorem states that a distributed data store can guarantee at most two of three properties — Consistency, Availability, and Partition Tolerance — when a network partition occurs.

Read definition

CDC (Change Data Capture)

CDC is the process of tracking and capturing row-level changes (inserts, updates, deletes) in a database and streaming those change events to downstream systems in near real-time.

Read definition

Checkpointing

Checkpointing is the process of flushing all committed changes from the write-ahead log (WAL) to the main database storage, and then clearing the…

Read definition

Circuit Breaker

A circuit breaker stops sending requests to a failing downstream service after consecutive failures, preventing cascading failures and giving the service time to recover.

Read definition

Clustering

In databases, clustering can refer to either grouping similar data together or deploying multiple nodes for scalability/HA. A database cluster is a set of…

Read definition

Columnar vs Row Storage

Columnar storage organizes data by columns (all values of one field together) while row storage groups complete records together, each optimized for different query patterns.

Read definition

Compaction

Compaction is the process of merging and recycling data files to reclaim space and improve read performance, especially in Log-Structured Merge (LSM) tree…

Read definition

Composite Index

A composite index (multi-column index) indexes multiple columns together in a single B-tree, optimizing queries that filter or sort on the combined column set while following the leftmost prefix rule.

Read definition

Connection Pooling

Connection pooling maintains a pool of open database connections that can be reused, reducing the overhead of opening/closing connections.

Read definition

Consensus

Consensus is the process by which distributed nodes agree on a single value or decision, even when some nodes fail, using algorithms like Raft and Paxos.

Read definition

Consistent Hashing

Consistent hashing is a technique to distribute keys across nodes so that minimal keys need to be moved when nodes join/leave.

Read definition

Covering Index

A covering index is a (secondary or composite) index that includes all columns needed for a query. The index “covers” the query, so the database can…

Read definition

D8

Data Lake

A data lake is a centralized repository that stores large volumes of raw data in its native format, allowing schema-on-read analysis with tools like Spark, Presto, and Athena.

Read definition

Data Locality

Data locality is the principle of placing computation (queries, processing) close to where the data resides, rather than moving large data sets across the…

Read definition

Data Migration

Data migration is the process of moving data between storage systems, formats, or schemas, typically during upgrades or platform changes.

Read definition

Data Skew

Data skew is an uneven distribution of data or workload across partitions or nodes, causing hotspots where some partitions handle disproportionate traffic or storage.

Read definition

Data Warehouse

A data warehouse is a centralized analytical store that holds curated, structured data optimized for complex queries and reporting, separate from operational systems.

Read definition

Deadlock

A deadlock occurs when two or more transactions permanently block each other, each holding a lock the other needs, creating a circular dependency that can only be resolved by aborting one transaction.

Read definition

Denormalization

Denormalization is the process of intentionally adding redundancy to a database schema to improve read/query performance. In a denormalized design, data…

Read definition

Dirty Read

A dirty read occurs when a transaction reads data written by another transaction that has not yet committed, risking exposure to data that may be rolled back.

Read definition

E2

Eventual Consistency

Eventual consistency guarantees that if no new updates are made, all replicas of a data item will converge to the same value over time, trading immediate consistency for higher availability and lower latency.

Read definition

Exactly-Once Semantics

Exactly-once semantics guarantees that a message or operation is processed once and only once, despite retries, crashes, and network failures.

Read definition

F2

Failover

Failover is the automatic or manual switching of operations from a failed primary system to a standby replica, ensuring service continuity with minimal downtime.

Read definition

Federation

Federation is an architecture where multiple autonomous databases or services coordinate to answer queries without sharing a single storage layer.

Read definition

G1

Geo-Replication

Geo-replication copies data across geographically distributed data centers to improve latency, availability, and disaster recovery.

Read definition

H4

Hash Partitioning

Hash partitioning distributes data across partitions by applying a hash function to a key, producing uniform distribution for point lookups.

Read definition

High Availability (HA)

High availability means a system is continuously operational with minimal downtime, typically measured as 99.9% to 99.999% uptime through redundancy, automatic failover, and distributed architecture.

Read definition

Horizontal Scaling

Horizontal scaling (scale-out) adds more machines to a system to handle increased load, distributing work across nodes rather than upgrading a single server.

Read definition

Hot Partition (Hot Shard)

A hot partition (or hot shard/key) is a data partition that receives disproportionately high traffic, causing resource saturation.

Read definition

I2

Idempotency

Idempotency means an operation produces the same result whether executed once or multiple times, making retries safe in distributed systems.

Read definition

Indexing

A database index is a data structure that improves the speed of data retrieval on a table at the cost of additional writes and storage. It works like a book’s…

Read definition

L5

Leader–Follower Replication

Leader–Follower (also called master-slave) replication is a strategy where one node (the leader) receives all write operations, and one or more follower…

Read definition

Load Balancing

Load balancing distributes incoming network requests across multiple servers to prevent overload, improve response times, and provide fault tolerance through redundancy.

Read definition

Lock Escalation

Lock escalation is the process of converting many fine-grained locks (like row or page locks) into a coarser lock (like a table lock) to reduce lock management…

Read definition

Logical vs Physical Replication

Logical replication streams row-level change events (inserts, updates, deletes) while physical replication copies raw WAL bytes, producing a byte-identical standby.

Read definition

LSM Tree

A Log-Structured Merge-tree (LSM tree) is a data structure optimized for high write volumes. It buffers writes in memory and periodically merges them to…

Read definition

M3

Materialized Views

A materialized view is a database object that stores the result of a query as a physical table. Unlike a regular (virtual) view, it contains actual…

Read definition

Multi-Leader Replication

Multi-leader (also called multi-master) replication allows multiple nodes (leaders) to accept writes. Data is replicated between leaders to keep them in…

Read definition

MVCC (Multi-Version Concurrency Control)

MVCC allows multiple transactions to read and write data concurrently without blocking each other, by maintaining multiple versions of each row and giving each transaction a consistent snapshot.

Read definition

N1

Normalization

Normalization is the process of organizing database schema to reduce redundancy and improve integrity. Typically, data is divided into multiple tables…

Read definition

O1

Optimistic Locking

Optimistic locking allows concurrent transactions to proceed without acquiring locks, detecting conflicts at commit time using version numbers or timestamps and retrying on collision.

Read definition

P2

Pessimistic Locking

Pessimistic locking acquires exclusive locks on data before accessing it, preventing concurrent modifications by blocking other transactions until the lock is released.

Read definition

Phantom Read

A phantom read occurs when a transaction re-executes a query with a range condition and gets a different set of rows because another transaction inserted or deleted matching rows in between.

Read definition

Q2

Query Planner (Cost-Based Optimizer)

The query planner/optimizer is a component that determines the most efficient way to execute a database query. A cost-based optimizer (CBO) estimates the…

Read definition

Quorum

In distributed systems, a quorum is the minimum number of votes (or nodes) that must agree to perform an operation. It is a technique to ensure consistency…

Read definition

R5

Range Partitioning

Range partitioning divides data into partitions based on ordered key ranges, keeping contiguous values together for efficient range scans.

Read definition

Read Replicas

A read replica is a read-only copy of a database instance. The primary database handles writes; read replicas asynchronously replicate data from the primary…

Read definition

Read Skew

Read skew is a concurrency anomaly where a transaction sees inconsistent data because it reads different rows or tables at different points in time during concurrent modifications.

Read definition

Rebalancing (Data Rebalancing)

Rebalancing is redistributing data across nodes or partitions to restore even load after cluster changes or skew.

Read definition

Resharding

Resharding is changing the number or configuration of shards (data partitions) in a sharded database, moving data between shards accordingly.

Read definition

S6

Schema Evolution

Schema evolution is the practice of modifying a data schema over time while maintaining compatibility with existing readers and writers.

Read definition

Schema Registry

A schema registry is a centralized service that stores, versions, and enforces compatibility of data schemas in event streaming and messaging systems.

Read definition

Secondary Indexes

A secondary index is any index on a table column that is not the primary key. It is stored separately from the table rows.

Read definition

Snapshot Isolation

Snapshot isolation is a transaction isolation level where each transaction operates on a snapshot of the database as of the start of the transaction.…

Read definition

Split-Brain

Split-brain occurs when a distributed cluster partitions into two or more segments that each believe they are the primary, leading to conflicting writes and data divergence.

Read definition

Strong Consistency

Strong consistency guarantees that every read returns the result of the most recent write, so all clients see the same data at the same time regardless of which node they query.

Read definition

T4

Three-Phase Commit (3PC)

Three-phase commit is an extension of 2PC designed to reduce blocking. It adds a “pre-commit” phase so participants can make a safe decision without indefinite…

Read definition

Throttling and Rate Limiting

Throttling controls resource usage by limiting the rate of requests. Rate limiting enforces a maximum number of requests in a time window to protect services from overload and abuse.

Read definition

Time-Series Partitioning

Time-series partitioning divides data into partitions by time intervals, optimizing for append-heavy workloads and time-windowed queries.

Read definition

Two-Phase Commit (2PC)

Two-phase commit is an atomic commitment protocol for distributed transactions. A coordinator asks all involved nodes (participants) whether they can commit.…

Read definition

V1

Vertical Scaling

Vertical scaling (scale-up) increases the capacity of a single machine by adding CPU, RAM, or faster storage, trading simplicity for a hardware ceiling.

Read definition

W2

Write Skew

Write skew is a concurrency anomaly where two transactions read overlapping data, then write to non-overlapping rows, producing a state that violates an application invariant neither transaction broke individually.

Read definition

Write-Ahead Logging (WAL)

Write-ahead logging (WAL) records all changes to a log file before applying them to the database, guaranteeing atomicity and durability by enabling crash recovery through log replay.

Read definition