hive-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Xiaomeng Huang (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HIVE-7934) Improve column level encryption with key management
Date Mon, 08 Dec 2014 11:26:12 GMT

    [ https://issues.apache.org/jira/browse/HIVE-7934?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14237779#comment-14237779
] 

Xiaomeng Huang commented on HIVE-7934:
--------------------------------------

Hi [~chirag.aggarwal]
When I worked on this feature, KMS feature is not released in hadoop 2.5.X. So I decided to
write a generic crypto codec and key management in Hive. But talked with some committer who
familiar with Security in hadoop-common, it is a few duplicated with the crypto codec in hadoop-common.
I just have a look at hadoop release notes, hadoop 2.6.0 seems like include KMS feature. HIVE-8049
is the initial patch to implement Hive Column Level Encrpytion based on KMS in hadoop-common.
And HIVE-8252 and HIVE-8416 will be closed as duplicated. I will update the patch of HIVE-8049
these days. Thanks for watching!

> Improve column level encryption with key management
> ---------------------------------------------------
>
>                 Key: HIVE-7934
>                 URL: https://issues.apache.org/jira/browse/HIVE-7934
>             Project: Hive
>          Issue Type: Improvement
>            Reporter: Xiaomeng Huang
>            Assignee: Xiaomeng Huang
>            Priority: Minor
>
> Now HIVE-6329 is a framework of column level encryption/decryption. But the implementation
in HIVE-6329 is just use Base64, it is not safe and have some problems:
> - Base64WriteOnly just be able to get the ciphertext from client for any users. 
> - Base64Rewriter just be able to get plaintext from client for any users.
> I have an improvement based on HIVE-6329 using key management via kms.
> # setup kms and set kms-acls.xml (e.g. user1 and root has permission to get key)
> {code}
>  <property>
>     <name>hadoop.kms.acl.GET</name>
>     <value>user1 root</value>
>     <description>
>       ACL for get-key-version and get-current-key operations.
>     </description>
>   </property>
> {code}
> # set hive-site.xml 
> {code}
>  <property>  
>     <name>hadoop.security.kms.uri</name>  
>     <value>http://localhost:16000/kms</value>  
>  </property> 
> {code}
> # create an encrypted table
> {code}
> -- region-aes-column.q
> drop table region_aes_column;
> create table region_aes_column (r_regionkey int, r_name string) ROW FORMAT SERDE 'org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe'
>   WITH SERDEPROPERTIES ('column.encode.columns'='r_name', 'column.encode.classname'='org.apache.hadoop.hive.serde2.aes.AESRewriter')
>   STORED AS TEXTFILE TBLPROPERTIES("hive.encrypt.keynames"="hive.k1");
> insert overwrite table region_aes_column
> select
>   r_regionkey, r_name
> from region;
> {code}
> # query table by different user, this is transparent to users. It is very convenient
and don't need to set anything.
> {code}
> [root@huang1 hive_data]# hive
> hive> select * from region_aes_column;
> OK
> 0	AFRICA
> 1	AMERICA
> 2	ASIA
> 3	EUROPE
> 4	MIDDLE EAST
> Time taken: 0.9 seconds, Fetched: 5 row(s)
> [root@huang1 hive_data]# su user1
> [user1@huang1 hive_data]$ hive
> hive> select * from region_aes_column;
> OK
> 0	AFRICA
> 1	AMERICA
> 2	ASIA
> 3	EUROPE
> 4	MIDDLE EAST
> Time taken: 0.899 seconds, Fetched: 5 row(s)
> [root@huang1 hive_data]# su user2
> [user2@huang1 hive_data]$ hive
> hive> select * from region_aes_column;
> OK
> 0	RcQycWVD
> 1	Rc8lam9Bxg==
> 2	RdEpeQ==
> 3	Qdcyd3ZH
> 4	ScskfGpHp8KIIuY=
> Time taken: 0.749 seconds, Fetched: 5 row(s)
> {code}



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message