What is Apache Cassandra ?

Cassandra is a distributed NoSql database, similar to Google BigTable or HBase.
It is hybrid between Key-Value and Column oriented database.
There are client drivers for all langages (internal protocol is binary “Thrift”).
There is also a langage CQL very similar to SQL, with an interactive shell to execute commands: CQLSH

Some links:

Google: Google Cassandra ... over 105 Millions results
Wikipedia: https://en.wikipedia.org/wiki/Apache_Cassandra
Home page: http://cassandra.apache.org/
Doc: Cassandra Doc

First steps starting a single Cassandra Server (2 minutes)

Download link: apache-cassandra-3.9-bin.tar.gz

Unzip, then start it:

apache-cassandra-3.9$ ./bin/cassandra -f
...
...
INFO  19:07:18 Starting listening for CQL clients on localhost/127.0.0.1:9042 (unencrypted)...
INFO  19:07:18 Not starting RPC server as requested. Use JMX (StorageService->startRPCServer()) or nodetool (enablethrift) to start it
INFO  19:07:27 Scheduling approximate time-check task with a precision of 10 milliseconds
INFO  19:07:28 Created default superuser role 'cassandra'

First steps using Cassandra CQL Shell (5 minutes)

Starting CQLSH shell command client:

$ bin/cqlsh localhost
Connected to Test Cluster at localhost:9042.
[cqlsh 5.0.1 | Cassandra 3.8 | CQL spec 3.4.2 | Native protocol v4]
Use HELP for help.

First command to check server topology (=single node)

cqlsh> SELECT cluster_name, listen_address FROM system.local;

 cluster_name | listen_address
--------------+----------------
 Test Cluster |      127.0.0.1

(1 rows)
cqlsh>

First commands to create Keyspace, then Table, then Data..

Cql DDL Doc

cqlsh> CREATE KEYSPACE myks WITH replication = {'class': 'SimpleStrategy', 'replication_factor' : 1};
cqlsh> 
cqlsh> USE myks;
cqlsh:myks>

Create table:

CREATE TABLE t1 (
    f1 int,
    f2 int,
    f3 text,
    f4 text,
    PRIMARY KEY (f1, f2)
);

CRUD DML commands:

INSERT INTO t1 (f1,f2,f3,f4) VALUES (0, 0, 'value0-0:f3', 'value0-0:f4');
INSERT INTO t1 (f1,f2,f3,f4) VALUES (0, 1, 'value0-1:f3', 'value0-1:f4');

SELECT * FROM t1;

It just works !

 f1 | f2 | f3          | f4
----+----+-------------+-------------
  0 |  0 | value0-0:f3 | value0-0:f4
  0 |  1 | value0-1:f3 | value0-1:f4

(2 rows)

Doing select with criteria

select * from t1 where f1 = 0 and f2 = 1;

 f1 | f2 | f3          | f4
----+----+-------------+-------------
  0 |  1 | value0-1:f3 | value0-1:f4

(1 rows)

Trying to do a select with un-indexed criteria … like “Full Table Scan” in sql:

cqlsh:myks> select * from t1 where f2 = 1;
InvalidRequest: Error from server: code=2200 [Invalid query] message="Cannot execute this query as it might involve data filtering and thus may have unpredictable performance. If you want to execute this query despite the performance unpredictability, use ALLOW FILTERING"

OK, let be explicitely doing wrong things like sql:

cqlsh:myks> select * from t1 where f2 = 1  ALLOW FILTERING;

 f1 | f2 | f3          | f4
----+----+-------------+-------------
  0 |  1 | value0-1:f3 | value0-1:f4

(1 rows)

Using secondary indexes:

cqlsh:myks> CREATE INDEX ON t1 (f3);

cqlsh:myks> select * from t1 where f3 = 'value0-0:f3';

 f1 | f2 | f3          | f4
----+----+-------------+-------------
  0 |  0 | value0-0:f3 | value0-0:f4

(1 rows)

Update 1 row using complete pk:

update t1 set f3 = 'f3' where f1=0 and f2=0;

Try update rows using partial pk:

cqlsh:myks> update t1 set f3 = 'f3' where f1=0;
InvalidRequest: Error from server: code=2200 [Invalid query] message="Some clustering keys are missing: f2"

Try update rows using non pk criteria:

cqlsh:myks> update t1 set f3 = 'f3' where f3 = 'value0-0:f3';
InvalidRequest: Error from server: code=2200 [Invalid query] message="Some partition key parts are missing: f1"

Preliminary conclusions on Cassandra & CQL : it looks really simple, really like SQL, but with “natural” limitations from a Key-Value database.

Using Java client (10 minutes)

I tryed to use the all-in-one springboot-data-cassandra, but as you can see later , it is in a broken state for Cassandra 3!!
Let’s start with a simple Cassandra driver connection.

Create a maven project, and edit your pom.xml to add

  <dependencies>
  	<dependency>
		<groupId>com.datastax.cassandra</groupId>
		<artifactId>cassandra-driver-core</artifactId>
		<version>3.1.3</version>
	</dependency>
  </dependencies>

Type “mvn install”, and import in eclipse.

First write java code to connect to Cassandra server:

import com.datastax.driver.core.Cluster;
import com.datastax.driver.core.Session;

	// Connect to the cluster and keyspace "myks"
	Cluster cluster = Cluster.builder().addContactPoint("127.0.0.1").build();
	Session session = cluster.connect("myks");
	System.out.println("connected to Cassandra localhost");

Then execute a query

import com.datastax.driver.core.ResultSet;
import com.datastax.driver.core.Row;

	// execute some queries...
	ResultSet results = session.execute("SELECT * FROM T1 WHERE f1=0");
	for (Row row : results) {
		int f1 = row.getInt("f1"); 
		int f2 = row.getInt("f2"); 
		String f3 = row.getString("f3"); 
		String f4 = row.getString("f4"); 
		System.out.println("row f1:" + f1 + ", f2:" + f2 + ", f3:'" + f3 + "', f4:'" + f4 + "'");
	}

It really looks like jdbc API, but simply replace java.sql.ResultSet by com.datastax.driver.core.ResultSet …

Run it … it just works!

connected to Cassandra localhost
row f1:0, f2:0, f3:'f3', f4:'value0-0:f4'
row f1:0, f2:1, f3:'value0-1:f3', f4:'value0-1:f4'

Using springboot-data-cassandra with cassandra (~ 2 hours trial solving errors)

Springboot is an amazing project, and springboot-data is also an amazing sub-project for simplifying java code for acessing repositories of data (not only jpa hibernate, but also key-values db, neo4j db, …).

To test a Java client using springboot-data + Cassandra, go to http://start.spring.io/
check option “Cassandra”, then click on “download” button
Unzip, and type “mvn install”, then in Eclipse “import as maven project”.
the project is only a maven skeleton, it does not contains code. If you ar einterrested, you can search for “google springdata cassandra sample”, or also read the doc http://docs.spring.io/autorepo/docs/spring-data-cassandra/current/reference/html/

I have then coded my class with @Table, @PrimaryKey, my primary key class with @PrimaryKeyClass and @PrimaryKeyColumn(..) columns and my interface Repository …

@PrimaryKeyClass
public class T1Key implements Serializable {

	/** */
	private static final long serialVersionUID = 1L;
	
	@PrimaryKeyColumn(ordinal = 0, type = PrimaryKeyType.PARTITIONED)
	private final int f1;

	@PrimaryKeyColumn(ordinal = 1, type = PrimaryKeyType.PARTITIONED)
	private final int f2;

	... 
	code ommitted : ctor + getters for f1, f2 + equals/hashCode
}


@Table
public class T1 {

	@PrimaryKey
	private T1Key pk;
	
	private String f3;
	private String f4;
	
	// ------------------------------------------------------------------------
	
	// silly constructor required by spring-data-cassandra WHY WHY?!! 
	// ... otherwise you have error "Property [pk] has no single column mapping"
	public T1() {
	}
	
	public T1(T1Key pk) {
		this.pk = pk;
	}

	// ------------------------------------------------------------------------

	// setter not needed on pk from spring-data-cassandra!
	public T1Key getPK() {
		return pk;
	}
	
	// dummy destructured getters getF1() and getF2() for composite key
	public int getF1() {
		return pk.getF1();
	}
	public int getF2() {
		return pk.getF2();
	}

	...
	code ommitted: regular getter/setters for field f3, f4
}


public interface T1Repository extends TypedIdCassandraRepository<T1,T1Key> {

	@Query("select * from T1 where f1=?0")
	public List<T1> findByF1(int f1);

}


@SpringBootApplication
public class TestCassandraApplication implements CommandLineRunner {

	@Autowired
	private T1Repository repository;

	@Value("${spring.data.cassandra.keyspace-name}")
	private String cassandraKeyspace;
	
	public static void main(String[] args) {
		SpringApplication.run(TestCassandraApplication.class, args);
	}

	@Override
	public void run(String... args) throws Exception {
		System.out.println("using Cassandra keyspace:" + cassandraKeyspace);
		
		long countAll = repository.count();
		System.out.println("count -> " + countAll + " elt(s)");
		Iterable<T1> all= repository.findAll();
		System.out.println("findAll -> [0]:" + all.iterator().next() + " ...");
		
		T1 res00 = repository.findOne(new T1Key(0, 0));
		System.out.println("find pk:(0,0) -> " + res00);
		
		List<T1> resLs = repository.findByF1(0);
		System.out.println("find by f1=0 -> " + resLs.size() + " elt(s)");
		
		int maxId = resLs.stream().mapToInt(x -> x.getPK().getF1()).max().getAsInt();
		T1Key newPK = new T1Key(maxId+1, 0);
		T1 newT1 = new T1(newPK);
		newT1.setF3("test from springboot-data");
		repository.save(newT1);
	}
}

The code with spring-data is very elegant. (Could be even better with lombok for getter/setter removal).

Unfortunatly, the first time I launched the client, I got this error ! … I forgot to change the schema keyspace ?

Caused by: com.datastax.driver.core.exceptions.NoHostAvailableException: All host(s) tried for query failed (tried: localhost/127.0.0.1:9042 (com.datastax.driver.core.exceptions.InvalidQueryException: unconfigured table schema_keyspaces))
	at com.datastax.driver.core.ControlConnection.reconnectInternal(ControlConnection.java:240) ~[cassandra-driver-core-2.1.9.jar:na]

The “magic” initialisation comes from class org.springframework.boot.autoconfigure.data.cassandra.CassandraDataAutoConfiguration.
It uses properties from org.springframework.boot.autoconfigure.cassandra.CassandraProperties

So I have added this line in file application.properties

spring.data.cassandra.keyspace-name: myks

… and re-executed “mvn install” !! (because eclipse does NOT update resources files from src/main/resources to target/classes)

But apparently, another problem comes from a mismatch between client (v2) and server (v3): Stackoverflow cassandra unconfigured table schema_keyspaces

From maven pom.xml dependency (mvn dependency:list, then mvn help:effective-pom), I discovered that the version used was cassandra-driver-core:2.1.9

I looked for an upgraded version of cassandra-driver-core.. using http://search.maven.org a:cassandra-driver-core I peeked up latest version 3.1.3

<properties>
	<cassandra-driver.version>3.1.3</cassandra-driver.version>
</properties>

Unfortunatly, this version does not compile, because cassadra-driver-dse is not released yet on same version 3.1.3, and version 3.0.0-rc1 is not compatible…

I got errors like, for an incompatibility between spring-data-cassandra and cassandra-driver …

Caused by: java.lang.NoSuchMethodError: com.datastax.driver.core.DataType.asJavaClass()Ljava/lang/Class;
	at org.springframework.data.cassandra.mapping.CassandraSimpleTypeHolder.<clinit>(CassandraSimpleTypeHolder.java:62) ~[spring-data-cassandra-1.4.6.RELEASE.jar:na]

.. as described here :StackOverflow spring-data-cassandra-1-3-4-not-compatible-with-cassandra-3

I tryed using the latest spring-data-cassandra, directly from github source code

git clone https://github.com/spring-projects/spring-data-cassandra.git
cd spring-data-cassandra
mvn clean install -DskipTests

even test do not compile! (so… rm -rf spring-data-cassandra/src/test/java/org)

I have also cloned the springboot-starter github project, and use the SNAPSHOT build..

git clone https://github.com/spring-projects/spring-boot.git
cd spring-boot
mvn clean install -DskipTests

	<parent>
		<groupId>org.springframework.boot</groupId>
		<artifactId>spring-boot-starter-parent</artifactId>
		<version>2.0.0.BUILD-SNAPSHOT</version>
		<relativePath/> <!-- lookup parent from repository -->
	</parent>

	<dependencyManagement>
		<dependencies>
			<dependency>
				<groupId>org.springframework.data</groupId>
				<artifactId>spring-data-cassandra</artifactId>
				<version>1.5.0.BUILD-SNAPSHOT</version>
			</dependency>
		</dependencies>
	</dependencyManagement>

Notice also that when not putting a default empty constructor on my @Table class, I had a silly confusing error message

Property [pk] has no single column mapping

Finally, I got some successful results!!

using Cassandra keyspace:myks
count -> 2 elt(s)
findAll -> [0]:fr.an.tests.T1@65d8dff8 ...
find pk:(0,0) -> fr.an.tests.T1@4730e0f0
find by f1=0 -> 2 elt(s)

Conclusion: waiting springboot-data to publish an official release that work successfully with Cassandra version 3 !!
Cassandra, springboot & springboot-data are amazing projects.