All MCQs

The NameNode tracks directories, file names, permissions, and block locations.
It does not normally store the actual user data blocks.

HDFS divides files into blocks and distributes them across DataNodes.
Replication provides resilience to node failure.

Hadoop provides HDFS for storage and frameworks such as MapReduce and YARN.
It is designed for clusters of commodity hardware.

Distributed execution introduces coordination, partitioning, serialization, and recovery concerns.
Efficient designs minimize communication and exploit parallelism.

Distributed networks can experience communication failures.
Partition-tolerant systems continue functioning while groups of nodes cannot communicate.

Availability requires the system to respond even when some nodes are unavailable.
The response may not contain the most recent data in some designs.

Consistency means all clients observe a single current value.
Under network partitions, systems may trade this against availability.

Warehouses organize curated data for consistent business analysis.
They commonly use schema-on-write and dimensional models.

Data lakes retain structured, semi-structured, and unstructured data.
Governance and metadata are needed to prevent them becoming disorganized.

Schema-on-write validates and organizes data during ingestion.
It supports predictable queries but is less flexible.

Schema-on-read keeps raw data flexible until analysis time.
It is common in data lakes and heterogeneous Big Data environments.

Replication improves availability and fault tolerance.
If one node fails, another copy can still be used.