卡桑德拉的独特计数 [英] Distinct count in Cassandra

问题描述

我在 cassandra中有一个chat_matches表。并希望获取所有消息from_id到具有行数（按to_id分组）的不同to_id的计数。

I have a table chat_matches in "cassandra". And want to get the count of all messages from_id to distinct to_id with count of rows(group by to_id).

CREATE TABLE chat_matches(
    chat_id uuid,
    from_id bigint,
    to_id bigint,
    message  text,
    message_send_time timestamp,
    PRIMARY KEY ((chat_id,from_id),message_send_time)
);

推荐答案

在卡桑德拉 count（* ）是一项非常昂贵的操作，需要扫描所有节点上的所有行，只是为了给您计数并可能产生超时异常。

In cassandra count(*) is a very costly operation, need to scan all the row from all the node just to give you the count and can generate timeout exception.

因此，与其使用 count（*），而是维护一个如下所示的计数器表：

So Instead of using count(*) maintain a counter table like the below one :

CREATE TABLE message_counter (
    from_id bigint,
    to_id bigint,
    count counter,
    primary key((from_id, to_id ))
);

出现新消息时，只需将count的值增加一即可。

When a new message appears just increment the value of count by one.

 UPDATE message_counter SET count = count + 1 WHERE from_id = 1 AND to_id = 2;

现在您可以非常有效地按from_id到to_id选择邮件计数组

Now you can select the message count group by from_id to to_id very efficiently

SELECT * FROM message_counter WHERE from_id = 1 AND to_id = 2;

这篇关于卡桑德拉的独特计数的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！