node.jscassandracassandra-driver

Is Cassandra schema-less?


I'm using a NodeJS Application to connect to Cassandra using cassandra-driver

After playing around with cassandra-driver, I noticed that we have define schema for each table before-hand. Could anyone please help me understand if Schema of every table needs to be defined in Cassandra before-hand?

Here's how I'm trying to connect to Cassandra DB from my nodeJS application:

const cassandra = require("cassandra-driver");
const client = new cassandra.Client({
 contactPoints: ['127.0.0.1'],
 localDataCenter: 'datacenter1',
 keyspace: 'testdb'
});

const executeCassandraQuery = async (query, params) => {
let fetchData;
try {
    fetchData = await client.execute(query, params, { prepare: false });
    console.log(`Successfully executed the query: ${JSON.stringify(fetchData)}`);
} catch (error) {
    console.log(`There was an error executing the query: ${error}`)
}
return fetchData;
}

client.connect(async function (err, data) {
if (err) {
    console.log(`There was an error connecting to Cassandra...${err}`)
}
else {
    console.log(`Connected to Cassandra...${data}`);
    const insertQuery = 'INSERT INTO testtable (id, firstname, lastname) VALUES(?,?,?)';
    const insertQueryParams = [1, "Tony", "Mathew"];

    const insertQueryResponse = await executeCassandraQuery(insertQuery, insertQueryParams);
    console.log(`insertQueryResponse : ${insertQueryResponse}`);
    const fetchQuery = 'SELECT * FROM employees';
}
});  

Before running the above nodeJS script, I have created the 'testdb' keyspace and 'testtable' table using the following cql commands:

CREATE KEYSPACE testdb WITH replication = {'class': 'SimpleStrategy', 'replication_factor' : 3};
CREATE TABLE testtable(id int PRIMARY KEY, name text);

Solution

  • No, it is not schema-less.

    Related answer applies:

    In Cassandra, how is it possible to save data in a column name while leaving the column value empty?

    Relevant text:

    No, not any more. This used to be possible. It required the old (pre-3.x storage engine) and use of a Thrift-based API. But tables built with CQL (and the new storage engine) require all columns to be defined-up front, and do not allow it at runtime (at least, not in the same way that Thrift did).