Fetching Cassandra row keys

Question

Assume a Cassandra datastore with 20 rows, with row keys named "r1" .. "r20".

Questions:

• How do I fetch the row keys of the first ten rows (r1 to r10)?

• How do I fetch the row keys of the next ten rows (r11 to r20)?

I'm looking for the Cassandra analogy to:

SELECT row_key FROM table LIMIT 0, 10;
SELECT row_key FROM table LIMIT 10, 10;

Answer 1

Take a look at:

list<KeySlice> get_range_slices(keyspace, column_parent, predicate, range, consistency_level)

Where your KeyRange tuple is (start_key, end_key) == (r1, r10)

Answer 2

Based on my tests there is no order for the rows (unlike columns). CQL 3.0.0 can retrieve row keys but not distinct (there should be a way that I do not know).I my case I do not know what my key range is, so I tried to retrieve all the keys with both Hector and Thrift, and sort the keys later. The performance test with CQL 3.0.0 for 100000 columns 200 rows was about 500 milliseconds, Hector around 100 and thrift about 50 milliseconds. My Row key here is integer. Hector code follows:

public void qureyRowkeys(){
    myCluster = HFactory.getOrCreateCluster(CLUSTER_NAME, "127.0.0.1:9160");
    ConfigurableConsistencyLevel ccl = new ConfigurableConsistencyLevel();
    ccl.setDefaultReadConsistencyLevel(HConsistencyLevel.ONE);
    myKeyspace = HFactory.createKeyspace(KEYSPACE_NAME, myCluster, ccl);
    RangeSlicesQuery<Integer, Composite, String> rangeSlicesQuery = HFactory.createRangeSlicesQuery(myKeyspace, IntegerSerializer.get(), 
            CompositeSerializer.get(), StringSerializer.get());
    long start = System.currentTimeMillis();
    QueryResult<OrderedRows<Integer, Composite, String>> result =
      rangeSlicesQuery.setColumnFamily(CF).setKeys(0, -1).setReturnKeysOnly().execute();
    OrderedRows<Integer, Composite, String> orderedRows = result.get();
    ArrayList<Integer> list = new ArrayList<Integer>();
    for(Row<Integer, Composite, String> row: orderedRows){
        list.add(row.getKey());
    }

    System.out.println((System.currentTimeMillis()-start));
    Collections.sort(list);
    for(Integer i: list){
        System.out.println(i);
    }
}

This is the Thrift code:

public void retreiveRows(){
    try {
        transport = new TFramedTransport(new TSocket("localhost", 9160));
        TProtocol protocol = new TBinaryProtocol(transport);
        client = new Cassandra.Client(protocol);
        transport.open();
        client.set_keyspace("prefdb");
        ColumnParent columnParent = new ColumnParent("events"); 
        SlicePredicate predicate = new SlicePredicate();
        predicate.setSlice_range(new SliceRange(ByteBuffer.wrap(new byte[0]), ByteBuffer.wrap(new byte[0]), false, 1));              
        KeyRange keyRange = new KeyRange();  //Get all keys
        keyRange.setStart_key(new byte[0]);
        keyRange.setEnd_key(new byte[0]);
        long start = System.currentTimeMillis();
        List<KeySlice> keySlices = client.get_range_slices(columnParent, predicate, keyRange, ConsistencyLevel.ONE);
        ArrayList<Integer> list = new ArrayList<Integer>();
        for (KeySlice ks : keySlices) {
                 list.add(ByteBuffer.wrap(ks.getKey()).getInt());
        }    
        Collections.sort(list);
        System.out.println((System.currentTimeMillis()-start));
        for(Integer i: list){
            System.out.println(i);
        }

        transport.close();
    } catch (Exception e) {
        e.printStackTrace();

    }
}