统计cassandra单表数据量

水深无声 2022-11-11 05:18 182阅读 0赞

当cassandra数据量很大时使用select count(\*)这种方式基本上是无法统计的，会返回如下类似错误信息：

Cassandra timeout during read query at Consitency ONE(1 responses were required but only 0 replica responed)

这时候可以借助[cassandra-count][]这个工具来实现count的统计，需要注意的是这个工具在工作时会对cassandra服务器CPU以及内存使用都会带来不同程度的压力，所以在线上尽量不要执行count操作，cassandra不适合做count统计，

1、下载cassandra-count工具，地址[https://github.com/brianmhess/cassandra-count][cassandra-count]

2、执行如下命令,数据量很大时可以通过调大numSplits值来避免read timeout问题

./cassandra-count -host xx.xx.xx.xx -keyspace ks -table table1 -numSplits 1024

PS：指令参考

Switch	Option	Default	Description
    -host	IP Address		Cassandra connection point - required.
    -keyspace	Keyspace Name		Cassandra keyspace - required.
    -table	Table Name		Cassandra table name - required.
    -configFile	Filename	none	Filename of configuration options
    -port	Port Number	9042	Cassandra native protocol port number
    -user	Username	none	Cassandra username
    -pw	Password	none	Cassandra password
    -ssl-truststore-path	Truststore Path	none	Path to SSL truststore
    -ssl-truststore-pwd	Truststore Password	none	Password to SSL truststore
    -ssl-keystore-path	Keystore Path	none	Path to SSL keystore
    -ssl-keystore-path	Keystore Password	none	Password to SSL keystore
    '-consistencyLevel	Consistency Level	LOCAL_ONE	CQL Consistency Level
    -numSplits	Number of Splits	Number of Token Ranges	Number of splits/queries to create
    -numFutures	Number of Futures	1000	Number of Java driver futures in flight.
    -splitSize	Size of Split in MB	16	Split size in MB
    -debug	Debug mode	0	Debug printing verbosity (0=none, 1=some, 2=verbose)

[cassandra-count]: https://github.com/brianmhess/cassandra-count