简体繁体 English

更新Hive表中的值

[英]Update a value in Hive Table

原文 2014-09-22 08:59:40 0 1 hadoop/ mapreduce/ hive

I have gone through some of the Stackoveflow question . 我已经解决了一些Stackoveflow问题。 but I was not able to do the same. 但我无法做到这一点。

Hive doesnot support update but a workaround for that is to create partition to table. Hive不支持更新，但是一种解决方法是创建表的分区。

How to update a record in Hive? 如何在Hive中更新记录？

1 个解决方案

The update feature in Hive is planned in the next release. Hive中的更新功能计划在下一个版本中发布。

As a workaround , You could try the below : 解决方法是，您可以尝试以下操作：

Add a flag column default as I and a timestamp column 添加标志列默认为I和时间戳列
consider partitions as your primary key fields (combination). 将分区视为您的主键字段（组合）。
whenever a new record (updated ) on this primary key combination set the flag as U. 每当此主键组合上的新记录（已更新）将标志设置为U时。
Write a custom serde class where in only U records are shown. 编写一个自定义serde类，其中仅显示U条记录。

NOTE : There will be duplicated data , but serde should only show the latest U data corresponding to the latest timestamp in the timestamp column. 注意：会有重复的数据，但serde应该仅在timestamp列中显示与最新时间戳相对应的最新U数据。