Page Comparison

...

For testing different scenarios keyspace and table created

CREATE KEYSPACE test WITH replication = {'class':'SimpleStrategy', 'replication_factor' : 2};

CREATE TABLE user(name text, id text, PRIMARY KEY (name));

In java

PreparedStatement preparedStatement = session.prepare("insert into test2test.user (id, name) values (?, ?)");
PreparedStatement preparedStatement2 = session.prepare("update test2test.user set id='id_updated-1' where name=?");
PreparedStatement preparedStatement3 = session.prepare("update test2test.user set id='id_updated-2' where name=?");


int i = 1;
while(i <= 1000) {
	batchStatement.add(preparedStatement.bind("id_"+i, "user-" + i));
	++i;
}

batchStatement.add(preparedStatement2.bind("user-1"));
batchStatement.add(preparedStatement3.bind("user-2"));

...

The above query will throw exception

Caused by: com.datastax.driver.core.exceptions.WriteTimeoutException: Cassandra timeout during write query at 
consistency ONEQUORUM (12 replica were required but only 01 acknowledged the write)

Observation

The 1000 row data got inserted into user table and got updated also as above batch execution.
The WriteTimeoutException.getWriteType() == BATCH, ensures data is inserted into the table

Approach

To handle above WriteTimeoutException we can use BatchStatement.Type

BatchStatement batchStatement = new BatchStatement(BatchStatement.Type.LOGGED);

Batch type of LOGGED, guarantees atomic insertion of data, and WriteTimeoutException.getWriteType() == BATCH, eventually written to the appropriate replicas and the developer doesn't have to do anything.

Thus, the above two checks handles partial write.