hbase 多版本version

╰半夏微凉° 2021-11-11 23:48 448阅读 0赞

# 欢迎关注本人公众号 #

![在这里插入图片描述][20200515113638671.jpg_pic_center]

# 默认保存version数量 #

测试SQL

create table PERSON3 (id integer not null primary key,name varchar,age integer, sex INTEGER, val DOUBLE);

创建成后，查看表结构信息：

hbase(main):018:0> describe 'PERSON3'
    Table PERSON3 is ENABLED                                                                                                                                                                       
    PERSON3, {TABLE_ATTRIBUTES => {coprocessor$1 => '|org.apache.phoenix.coprocessor.ScanRegionObserver|805306366|', coprocessor$2 => '|org.apache.phoenix.coprocessor.UngroupedAggregateRegionObse
    rver|805306366|', coprocessor$3 => '|org.apache.phoenix.coprocessor.GroupedAggregateRegionObserver|805306366|', coprocessor$4 => '|org.apache.phoenix.coprocessor.ServerCachingEndpointImpl|805
    306366|', coprocessor$5 => '|org.apache.phoenix.hbase.index.Indexer|805306366|org.apache.hadoop.hbase.index.codec.class=org.apache.phoenix.index.PhoenixIndexCodec,index.builder=org.apache.pho
    enix.index.PhoenixIndexBuilder', coprocessor$6 => '|org.apache.hadoop.hbase.regionserver.LocalIndexSplitter|805306366|'}                                                                       
    COLUMN FAMILIES DESCRIPTION                                                                                                                                                                    
    {NAME => '0', VERSIONS => '1', EVICT_BLOCKS_ON_CLOSE => 'false', NEW_VERSION_BEHAVIOR => 'false', KEEP_DELETED_CELLS => 'FALSE', CACHE_DATA_ON_WRITE => 'false', DATA_BLOCK_ENCODING => 'FAST_D
    IFF', TTL => 'FOREVER', MIN_VERSIONS => '0', REPLICATION_SCOPE => '0', BLOOMFILTER => 'ROW', CACHE_INDEX_ON_WRITE => 'false', IN_MEMORY => 'false', CACHE_BLOOMS_ON_WRITE => 'false', PREFETCH_
    BLOCKS_ON_OPEN => 'false', COMPRESSION => 'NONE', BLOCKCACHE => 'true', BLOCKSIZE => '65536'}

可以看到`VERSIONS => '1'` 。默认是1，只保留一个版本，也就是说如果有数据写入，那么较老的时间戳的数据会忽略作废。  
下面验证上面的观点。  
首先更改默认保留的version数量,将0列族的versions改为5：

hbase(main):017:0> alter 'PERSON3' ,{NAME=>'0', VERSIONS=>5}
    Updating all regions with the new schema...
    1/1 regions updated.
    ^[[ADone.
    Took 2.5956 seconds                                                                                                                                                                            
    hbase(main):018:0> describe 'PERSON3'
    Table PERSON3 is ENABLED                                                                                                                                                                       
    PERSON3, {TABLE_ATTRIBUTES => {coprocessor$1 => '|org.apache.phoenix.coprocessor.ScanRegionObserver|805306366|', coprocessor$2 => '|org.apache.phoenix.coprocessor.UngroupedAggregateRegionObse
    rver|805306366|', coprocessor$3 => '|org.apache.phoenix.coprocessor.GroupedAggregateRegionObserver|805306366|', coprocessor$4 => '|org.apache.phoenix.coprocessor.ServerCachingEndpointImpl|805
    306366|', coprocessor$5 => '|org.apache.phoenix.hbase.index.Indexer|805306366|org.apache.hadoop.hbase.index.codec.class=org.apache.phoenix.index.PhoenixIndexCodec,index.builder=org.apache.pho
    enix.index.PhoenixIndexBuilder', coprocessor$6 => '|org.apache.hadoop.hbase.regionserver.LocalIndexSplitter|805306366|'}                                                                       
    COLUMN FAMILIES DESCRIPTION                                                                                                                                                                    
    {NAME => '0', VERSIONS => '5', EVICT_BLOCKS_ON_CLOSE => 'false', NEW_VERSION_BEHAVIOR => 'false', KEEP_DELETED_CELLS => 'FALSE', CACHE_DATA_ON_WRITE => 'false', DATA_BLOCK_ENCODING => 'FAST_D
    IFF', TTL => 'FOREVER', MIN_VERSIONS => '0', REPLICATION_SCOPE => '0', BLOOMFILTER => 'ROW', CACHE_INDEX_ON_WRITE => 'false', IN_MEMORY => 'false', CACHE_BLOOMS_ON_WRITE => 'false', PREFETCH_
    BLOCKS_ON_OPEN => 'false', COMPRESSION => 'NONE', BLOCKCACHE => 'true', BLOCKSIZE => '65536'}

put命令可以指定时间戳，也可以不指定。不指定默认是hbase集群服务器的时间。  
下面put几条数据

hbase(main):019:0> put 'PERSON3' , 'row1', '0:a','3'
    Took 0.0062 seconds                                                                                                                                                                            
    hbase(main):020:0> scan 'PERSON3'
    ROW                                              COLUMN+CELL                                                                                                                                   
     row1                                            column=0:a, timestamp=1565005841680, value=1                                                                                                  
     
    hbase(main):021:0> scan 'PERSON3', {VERSIONS => 10}
    ROW                                              COLUMN+CELL                                                                                                                                   
     row1                                            column=0:a, timestamp=1565005841680, value=1                                                                                                  
     row1                                            column=0:a, timestamp=1564971755759, value=a                                                                                                  
     row1                                            column=0:a, timestamp=1564826176269, value=3

接下来指定时间戳插入1条数据：

hbase(main):022:0> put 'PERSON3' , 'row1', '0:a','4', 1564826176000
    Took 0.0038 seconds                                                                                                                                                                            
    hbase(main):023:0> scan 'PERSON3', {VERSIONS => 10}
    ROW                                              COLUMN+CELL                                                                                                                                   
     row1                                            column=0:a, timestamp=1565005841680, value=1                                                                                                  
     row1                                            column=0:a, timestamp=1564971755759, value=a                                                                                                  
     row1                                            column=0:a, timestamp=1564826176269, value=3                                                                                                  
     row1                                            column=0:a, timestamp=1564826176000, value=4

由于手动指定的1564826176000比现有数据小，所以会排在末尾。查询时，默认返回最新的一条记录，1564826176000不会被查询出来。

hbase(main):025:0> get 'PERSON3', 'row1','0:a'
    COLUMN                                           CELL                                                                                                                                          
     0:a                                             timestamp=1565005841680, value=1                                                                                                              
    1 row(s)

可以利用hbase的这种机制来解决实际的问题。比如在hbase集群数据同步，不管是通过hlog还是kafka数据传输，都需要保证数据的顺序性。那么，从集群A同步到集群B的场景中，A过来的数据添加上时间戳，插入集群B时指定该时间戳，就可以保证在B集群已经有新数据put进来时，A集群的数据不会在最新的位置，保证了数据的顺序性。

参考资料：https://blog.csdn.net/ck3207/article/details/81674557

[20200515113638671.jpg_pic_center]: /images/20211109/8ed0979a93724f9abcab87253822f12e.png