Apache Cassandra, 3rd Edition - Part 2

Question 1

What is Cassandra’s data model called?

Answer 1

A partitioned row store - data is stored in sparse multidimensional hash tables where each row has a unique partition key used to distribute rows across nodes.

Question 2

What are the basic Cassandra data structures in order from smallest to largest?

Answer 2

Column (name/value pair) → Row (container for columns) → Partition (group of related rows) → Table (container for rows) → Keyspace (container for tables) → Cluster (container for keyspaces)

Question 3

What is a keyspace in Cassandra?

Answer 3

The outermost container for data, corresponding to a database in relational terms. Has attributes defining keyspace-wide behavior like replication.

Question 4

What is a partition in Cassandra?

Answer 4

A group of related rows stored together on the same nodes. Each partition is uniquely identified by a partition key.

Question 5

What is a partition key?

Answer 5

The column(s) used to determine which node stores the data. It’s hashed by the partitioner to assign data to nodes in the ring.

Question 6

What is a clustering column?

Answer 6

Columns that control how data is sorted within a partition. Combined with partition key, they form the primary key and uniquely identify rows.

Question 7

What is a primary key in Cassandra?

Answer 7

Combination of partition key and optional clustering columns. The partition key determines data distribution; clustering columns determine sort order within partitions.

Question 8

How do you define a composite partition key?

Answer 8

Surround multiple columns with parentheses: PRIMARY KEY ((col1, col2), col3). The columns in parentheses form the partition key.

Question 9

What is a static column?

Answer 9

A column that stores data shared by every row in a partition. Defined with STATIC keyword. Not part of the primary key but shared across the partition.

Question 10

What is the text data type?

Answer 10

A UTF-8 character string. Synonym for varchar. Recommended over ascii for internationalization support.

Question 11

What is the uuid data type?

Answer 11

A Type 4 UUID (128-bit) based entirely on random numbers. Often used as surrogate keys. Generate with uuid() function.

Question 12

What is the timeuuid data type?

Answer 12

A Type 1 UUID based on MAC address, system time, and a sequence number. Useful for conflict-free timestamps. Functions: now(), dateOf(), unixTimestampOf().

Question 13

What is the counter data type?

Answer 13

A 64-bit signed integer that can only be incremented or decremented, not set directly. Used for tracking statistics like page views.

Question 14

What are the restrictions on counter columns?

Answer 14

Cannot be part of primary key. If a table has a counter, all non-primary key columns must be counters. Counter operations are not idempotent.

Question 15

What is a set collection in Cassandra?

Answer 15

Stores unordered collection of unique elements. Returned in sorted order. Can add/remove items without reading first. Syntax: set<text></text>

Question 16

What is a list collection in Cassandra?

Answer 16

Stores ordered collection of elements. Elements accessed by position (index). Can prepend/append items. Updating by index can be expensive. Syntax: list<text></text>

Question 17

What is a map collection in Cassandra?

Answer 17

Stores collection of key-value pairs. Keys and values can be any type except counter. Access individual items by key. Syntax: map<text frozen<address>>

Question 18

What is a tuple in Cassandra?

Answer 18

A fixed-length set of values of various types. Must update entire tuple at once (no individual field updates). Less commonly used than UDTs. Syntax: tuple<text, text, int>

Question 19

What is a User-Defined Type (UDT)?

Answer 19

Custom types to extend the data model. Easier to use than tuples since values are named. Scoped by keyspace. Created with CREATE TYPE command.

Question 20

What does FROZEN mean for collections?

Answer 20

Freezing serializes a collection as a single binary blob. Required for nesting collections. Frozen collections must be read/written in entirety.

Question 21

What is the TTL (Time to Live)?

Answer 21

Column-level expiration. After TTL seconds, Cassandra marks data for deletion. Set with USING TTL clause. Check with TTL() function.

Question 22

How do you set TTL on data?

Answer 22

INSERT INTO table (cols) VALUES (vals) USING TTL 86400; or UPDATE table USING TTL 3600 SET col=val WHERE… (86400 = 24 hours)

Question 23

What is the timestamp of a column?

Answer 23

Every column stores a timestamp of when it was last modified. Used for conflict resolution (last write wins). View with writetime() function.

Question 24

What is the blob data type?

Answer 24

Binary large object - arbitrary array of bytes. Useful for media or binary files. Cassandra doesn’t validate blob contents. Represented as hex digits.

Question 25

What is the inet data type?

Answer 25

Represents IPv4 or IPv6 internet addresses. IPv4 displayed in dotted decimal format. IPv6 in eight groups of four hex digits.

Question 26

What is the boolean data type?

Answer 26

Simple true/false value. cqlsh is case insensitive for input but outputs True or False with capital letters.

Question 27

What is the difference between int and bigint?

Answer 27

int is a 32-bit signed integer. bigint is a 64-bit signed long integer (equivalent to Java long).

Question 28

What are smallint and tinyint?

Answer 28

smallint is a 16-bit signed integer. tinyint is an 8-bit signed integer. Both added in Cassandra 2.2.

Question 29

What is the varint data type?

Answer 29

Variable precision signed integer equivalent to java.math.BigInteger. Useful when you need arbitrary precision.

Question 30

What is the decimal data type?

Answer 30

Variable precision decimal equivalent to java.math.BigDecimal. Useful for financial calculations requiring exact precision.

Question 31

What is the date data type?

Answer 31

Represents a date without time of day. Added in Cassandra 2.2. Supports ISO 8601 formats. Stored as days since epoch.

Question 32

What is the time data type?

Answer 32

Represents time of day without a date. Added in Cassandra 2.2. Stored as nanoseconds since midnight.

Question 33

What is the timestamp data type?

Answer 33

64-bit signed integer representing both date and time. Supports ISO 8601 formats. Best practice: always include timezone.

Question 34

Why can't you modify a primary key after table creation?

Answer 34

The primary key controls how data is distributed in the cluster and stored on disk. Changing it would require reorganizing all data.

Question 35

What is the CREATE KEYSPACE syntax?

Answer 35

CREATE KEYSPACE name WITH replication = {'class': 'NetworkTopologyStrategy', 'dc1': 3, 'dc2': 2};

Question 36

What is the CREATE TABLE syntax?

Answer 36

CREATE TABLE keyspace.table (col1 type, col2 type, PRIMARY KEY ((partition_cols), clustering_cols)) WITH options;

Question 37

How do you add a column to an existing table?

Answer 37

ALTER TABLE table_name ADD column_name type;

Question 38

How do you drop a column from a table?

Answer 38

ALTER TABLE table_name DROP column_name;

Question 39

What is the INSERT syntax in CQL?

Answer 39

INSERT INTO table (col1, col2) VALUES (val1, val2) [USING TTL seconds] [AND TIMESTAMP microseconds];

Question 40

What is the UPDATE syntax in CQL?

Answer 40

UPDATE table [USING TTL seconds] SET col1 = val1 WHERE partition_key = value AND clustering_col = value;

Question 41

What is an UPSERT in Cassandra?

Answer 41

INSERT and UPDATE are semantically identical - both create a row if it doesn't exist or update if it does. There's no separate insert-only operation.

Question 42

How do you delete data in CQL?

Answer 42

DELETE FROM table WHERE partition_key = value; or DELETE column FROM table WHERE... to delete specific columns.

Question 43

What is TRUNCATE in CQL?

Answer 43

TRUNCATE table_name; removes all data from a table but keeps the table structure. Creates snapshots by default before truncating.

Question 44

How do you create a secondary index?

Answer 44

CREATE INDEX [name] ON table (column); Name defaults to tablename_columnname_idx if not specified.

Question 45

When should you avoid secondary indexes?

Answer 45

1) High cardinality columns 2) Very low cardinality columns 3) Frequently updated/deleted columns. Denormalization or materialized views often perform better.

Question 46

What is SASI (SSTable Attached Secondary Index)?

Answer 46

An alternative secondary index implementation where indexes are stored as part of SSTable files. Supports inequality searches and LIKE pattern matching.

Question 47

How do you create a SASI index?

Answer 47

CREATE CUSTOM INDEX name ON table (column) USING 'org.apache.cassandra.index.sasi.SASIIndex' WITH OPTIONS = {'mode': 'CONTAINS'};

Question 48

What is a materialized view?

Answer 48

A view that stores preconfigured query results, maintained automatically by Cassandra when base table changes. Supports queries on columns not in original primary key.

Question 49

What are the restrictions on materialized views?

Answer 49

Primary key must include all columns from base table's primary key. Only one non-primary key column from base can be added. Still experimental as of 4.0.

Question 50

How do you create a materialized view?

Answer 50

CREATE MATERIALIZED VIEW view_name AS SELECT * FROM table WHERE col IS NOT NULL PRIMARY KEY (new_pk_cols);

Question 51

What is the ALLOW FILTERING clause?

Answer 51

Overrides Cassandra's default behavior to allow queries that may require scanning all nodes. Should be avoided in production due to unpredictable performance.