Skip to content

Latest commit

 

History

History

cassandra

Cassandra implementation of Broker interface

The API was tested with Cassandra 3 and supports:

  • UDT (User Defined Types) / embedded structs and honors gocql.Marshaler/gocql.Unmarshaler
  • handling all primitive types (like int aliases , IP address);
    • net.IP can be stored as ipnet
    • net.IPNet can be stored with a MarshalCQL/UnmarshalCQL wrapper go structure
  • dumping all rows except for Table
  • quering by secondary indexes
  • mocking of gocql behavior (using gockle library)in automated unit tests

Cassandra Timeouts

The API will allow the client to configure either single node or multi-node cluster. Also, the client can configure following timeouts:

  • DialTimeout
    • Initial connection timeout, used during initial dial to server
    • Default value is 600ms
  • OpTimeout
    • Connection timeout, used during executing query
    • Default value is 600ms
  • RedialInterval
    • If not zero, gocql attempt to reconnect known DOWN nodes in every ReconnectSleep (ReconnectInterval)
    • Default value is 60s
  • Example
    config := &cassandra.Config{
        Endpoints:      127.0.0.1,
        Port:           9042,
        DialTimeout:    600 * time.Millisecond,
        OpTimeout:      600 * time.Millisecond,
        RedialInterval: 60 * time.Second,
    }

    clientConfig, err := cassandra.ConfigToClientConfig(config)

The timeout parameters are defined here config.go

Supported by underlying gocql structure ClusterConfig

Cassandra Data Consistency

The API will allow the client to configure consistency level for both

  • Session
  • Query (to be implemented)

Supported by underlying gocql structure Session

Factors to be considered for achieving desired consistency level.

  • Replication strategy

    • A replication strategy determines the nodes where replicas are placed.
      • SimpleStrategy: Use for a single data center only.
      • NetworkTopologyStrategy: Used for more than one data center.
  • Replication factor

    • The total number of replicas across the cluster is referred to as the replication factor.
    • A replication factor of 1 means that there is only one copy of each row on one node.
    • A replication factor of 2 means that there are two copies of each row, and each copy is stored on a different node.
    • The replication factor should not exceed the number of nodes in the cluster.
  • To achieve Quorum

    • Quorum = (sum_of_replication_factors / 2) + 1
    • (nodes_written + nodes_read) > replication_factor
  • References