Wide Column Stores (Cassandra, HBase)

Question 1

How does Cassandra achieve high availability?

Answer 1

Cassandra uses peer-to-peer architecture, replication across nodes, and tunable consistency levels to ensure high availability even during node failures.

Question 2

Explain the data model of Cassandra.

Answer 2

Cassandra’s data model is based on a wide-column structure with tables, rows, and columns grouped by partitions and clustering keys.

Question 3

What workloads is Cassandra optimized for?

Answer 3

Cassandra is optimized for high-throughput, write-heavy, distributed workloads with linear scalability and high availability.

Question 4

What is the column-family structure in Cassandra?

Answer 4

A column family is similar to a table, where data is organized into rows grouped by partition keys, and within each partition, sorted by clustering columns.

Question 5

What is a partition key in Cassandra?

Answer 5

A partition key determines the node where the data is stored using consistent hashing, and groups rows into partitions.

Question 6

What are clustering columns in Cassandra?

Answer 6

Clustering columns define how data is sorted within a partition and support range queries on related rows.

Question 7

What is CQL (Cassandra Query Language)?

Answer 7

CQL is a SQL-like language used to interact with Cassandra, including creating tables, inserting data, and querying.

Question 8

What is consistent hashing in Cassandra?

Answer 8

Consistent hashing is used to distribute data across nodes evenly by assigning token ranges to nodes.

Question 9

What is tunable consistency in Cassandra?

Answer 9

Cassandra allows clients to choose consistency levels like ONE, QUORUM, or ALL for reads and writes, balancing latency and data accuracy.

Question 10

What is QUORUM in Cassandra consistency?

Answer 10

QUORUM requires a majority of replicas to respond, offering a balance between consistency and availability.

Question 11

What does consistency level ONE mean in Cassandra?

Answer 11

ONE means only one replica must respond, favoring availability and speed over consistency.

Question 12

What does consistency level ALL mean in Cassandra?

Answer 12

ALL requires all replicas to acknowledge the read/write, ensuring strong consistency but reducing availability.

Question 13

What is replication in Cassandra?

Answer 13

Replication refers to storing multiple copies of data across different nodes to ensure redundancy and fault tolerance.

Question 14

What is replication factor in Cassandra?

Answer 14

The replication factor determines how many copies of data are stored; e.g., a factor of 3 means 3 copies.

Question 15

What is the write path in Cassandra?

Answer 15

Writes go to the commit log, then to an in-memory Memtable, and are eventually flushed to SSTables on disk.

Question 16

What is the read path in Cassandra?

Answer 16

Reads involve checking the Memtable, Bloom filters, SSTables, and potentially multiple replicas based on consistency level.

Question 17

What is HBase?

Answer 17

HBase is an open-source, distributed, scalable wide-column store built on top of HDFS for real-time read/write access to large datasets.

Question 18

What are RegionServers in HBase?

Answer 18

RegionServers handle read and write requests for regions (subsets of tables), store HFiles and MemStores, and manage compactions.

Question 19

What is the HMaster in HBase?

Answer 19

HMaster coordinates RegionServers, handles schema changes, and manages region assignment and load balancing.

Question 20

What are use cases for Cassandra?

Answer 20

Common use cases include time-series data, IoT, messaging systems, recommendation engines, and real-time analytics.

Question 21

What are use cases for HBase?

Answer 21

HBase is used in data lakes, batch and real-time analytics, time-series storage, and storing sparse datasets.

Question 22

What are wide-column stores?

Answer 22

Wide-column stores are NoSQL databases that store data in tables, rows, and dynamic columns grouped by column families (e.g., Cassandra, HBase).

Question 23

What are advantages of wide-column stores?

Answer 23

Advantages include high write throughput, horizontal scalability, flexible schemas, and efficient handling of sparse data.

Question 24

What are disadvantages of wide-column stores?

Answer 24

Disadvantages include complex data modeling, limited ad-hoc querying, and higher learning curve compared to relational databases.

Question 25

What are best practices for Cassandra schema design?

Answer 25

Model based on queries, use composite keys wisely, avoid large partitions, and keep write and read paths efficient.

Question 26

What are best practices for HBase usage?

Answer 26

Design row keys for even distribution, monitor region splits, use bulk loading for batch inserts, and leverage TTLs for cleanup.

Question 27

What are architectural implications of using Cassandra?

Answer 27

Cassandra enables decentralized architecture with no single point of failure, supporting global distribution and eventual consistency.

Question 28

What are architectural implications of using HBase?

Answer 28

HBase relies on Hadoop’s HDFS for storage and uses a master-slave model, requiring integration with HDFS and Zookeeper.

Question 29

How do wide-column stores perform under load?

Answer 29

They scale horizontally with high availability and write throughput, especially when using proper partitioning and compaction strategies.

Question 30

How do Cassandra and HBase provide fault tolerance?

Answer 30

Cassandra uses replication and hinted handoff; HBase uses HDFS replication and RegionServer failover managed by HMaster.

Question 31

How can you monitor Cassandra?

Answer 31

Tools like Prometheus, Grafana, DataStax OpsCenter, and `nodetool` are used to monitor latency, disk usage, and node health.

Question 32

How can you monitor HBase?

Answer 32

HBase metrics are available via JMX and can be visualized using Grafana, Ganglia, or Cloudera Manager.

Question 33

How do you debug performance issues in Cassandra?

Answer 33

Use tracing, slow query logs, `nodetool` stats, and check for hotspots, tombstones, and GC pauses.

Question 34

How do you debug HBase issues?

Answer 34

Check region logs, use HBase UI, review HDFS and ZooKeeper health, and analyze compaction and GC logs.

Question 35

What are real-world tradeoffs of using Cassandra?

Answer 35

Cassandra offers high availability and scalability but trades off immediate consistency and requires careful data modeling.

Question 36

What are real-world tradeoffs of using HBase?

Answer 36

HBase provides strong consistency and Hadoop integration but has operational complexity and less flexibility in querying.

Question 37

What are common Cassandra interview questions?

Answer 37

Examples: 'Explain tunable consistency', 'What is a partition key?', 'Describe the write path', 'How does Cassandra handle replication?'

Question 38

What are common HBase interview questions?

Answer 38

Examples: 'What is a RegionServer?', 'How does HBase store data?', 'Describe HMaster role', 'Explain compaction in HBase.'

Question 39

What are potential gotchas in Cassandra?

Answer 39

Hot partitions, large tombstones, unbounded row growth, and misconfigured consistency levels can cause performance issues.

Question 40

What are potential gotchas in HBase?

Answer 40

Region hotspots, unbalanced region splits, improper row key design, and compaction storms can affect performance.