Default Load Balancing Policy

By default, Ruby driver will use a combination of Token aware and Data Center aware round robin policies for load balancing.

This combination proved to be the most performant of of all built-in load balancing policies.

When the name of the local data center is not specified explicitly using Cassandra.cluster, the first datacenter seen by the load balancing policy will be considered local. Therefore, care must be taken to only include addresses of the nodes in the same datacenter as the application using the Ruby Driver in the :hosts option to Cassandra.cluster, or to provide :datacenter option explicitly.

Background

Given

a running cassandra cluster in 2 datacenters with 2 nodes in each

And

the following schema:

CREATE KEYSPACE simplex WITH replication = {'class': 'NetworkTopologyStrategy', 'dc1': '1', 'dc2': '1'};
USE simplex;
CREATE TABLE songs (
  id uuid PRIMARY KEY,
  title text,
  album text,
  artist text,
  tags set<text>,
  data blob
);
INSERT INTO songs (id, title, album, artist, tags)
VALUES (
   756716f7-2e54-4715-9f00-91dcbea6cf50,
   'La Petite Tonkinoise',
   'Bye Bye Blackbird',
   'Joséphine Baker',
   {'jazz', '2013'})
;
INSERT INTO songs (id, title, album, artist, tags)
VALUES (
   f6071e72-48ec-4fcb-bf3e-379c8a696488,
   'Die Mösch',
   'In Gold',
   'Willi Ostermann',
   {'kölsch', '1996', 'birds'}
);
INSERT INTO songs (id, title, album, artist, tags)
VALUES (
   fbdf82ed-0063-4796-9c7c-a3d4f47b4b25,
   'Memo From Turner',
   'Performance',
   'Mick Jager',
   {'soundtrack', '1991'}
);

Default load balancing policy always routes to primary replicas when possible

Given

the following example:

require 'cassandra'

cluster   = Cassandra.cluster(hosts: ['127.0.0.1', '127.0.0.2'])
session   = cluster.connect('simplex')
statement = session.prepare("SELECT token(id) FROM songs WHERE id = ?")

coordinator_ips = 4.times.map do
  info = session.execute(statement, arguments: [Cassandra::Uuid.new('756716f7-2e54-4715-9f00-91dcbea6cf50')]).execution_info
  info.hosts.last.ip
end

puts coordinator_ips.sort.uniq

When

it is executed

Then

its output should contain:

127.0.0.2

Default load balancing policy always uses primary replicas from the local datacenter

Given

the following example:

require 'cassandra'

cluster   = Cassandra.cluster(hosts: ['127.0.0.3', '127.0.0.4'])
session   = cluster.connect('simplex')
statement = session.prepare("SELECT token(id) FROM songs WHERE id = ?")

coordinator_ips = 4.times.map do
  info = session.execute(statement, arguments: [Cassandra::Uuid.new('756716f7-2e54-4715-9f00-91dcbea6cf50')]).execution_info
  info.hosts.last.ip
end

puts coordinator_ips.sort.uniq

When

it is executed

Then

its output should contain:

127.0.0.4

Default load balancing allows specifying data center explicitly

Given

the following example:

require 'cassandra'

cluster   = Cassandra.cluster(
              datacenter: 'dc1',
              hosts: ['127.0.0.3', '127.0.0.4']
            )
session   = cluster.connect('simplex')
statement = session.prepare("SELECT token(id) FROM songs WHERE id = ?")

coordinator_ips = 4.times.map do
  info = session.execute(statement, arguments: [Cassandra::Uuid.new('756716f7-2e54-4715-9f00-91dcbea6cf50')]).execution_info
  info.hosts.last.ip
end

puts coordinator_ips.sort.uniq

When

it is executed

Then

its output should contain:

127.0.0.2

Default load balancing policy prevents requests to remote datacenters

Given

the following example:

require 'cassandra'

cluster   = Cassandra.cluster(
              hosts: ['127.0.0.1', '127.0.0.2']
            )
session   = cluster.connect('simplex')
statement = "SELECT token(id) FROM songs"

$stdout.puts("=== START ===")
$stdout.flush

$stdin.gets # ready, block on stdin

begin
  execution_info = session.execute(statement).execution_info
  $stdout.puts("Statement #{statement.inspect} fulfilled by #{execution_info.hosts.last.ip}")
rescue => e
  $stdout.puts "#{e.class.name}: #{e.message}"
end
$stdout.flush
$stdout.puts("=== STOP ===")
$stdout.flush

And

it is running interactively

And

I wait for its output to contain “START”

When

node 1 stops

And

node 2 stops

And

I close the stdin stream

Then

its output should contain:

Cassandra::Errors::NoHostsAvailable: All hosts down