引言
hbase 提供很方便的 shell 腳本，可以對數據表進行 CURD 操作，但是畢竟是有一定的學習成本的，基本上對於開發來講，sql 語句都是看家本領，那么，有沒有一種方法可以把 sql 語句轉換成 hbase的原生API呢？ 這樣就可以通過普通平常的 sql 來對hbase 進行數據的管理，使用成本大大降低。
Apache Phoenix 組件就完成了這種需求，官方注解為 “Phoenix - we put the SQL back in NoSql”，通過官方說明，Phoenix 的性能很高，相對於 hbase 原生的scan 並不會差多少，而對於類似的組件 hive、Impala等，性能有着顯著的提升，詳細請閱讀 https://phoenix.apache.org/performance.html。
Apache Phoenix 官方站點：https://phoenix.apache.org/
Phoenix支持的sql語句： https://phoenix.apache.org/language/index.html
Phoenix 支持的DataTypes：https://phoenix.apache.org/language/datatypes.html
Phoenix 支持的函數：https://phoenix.apache.org/language/functions.html
安裝使用
Phoenix 安裝很簡單，下載對應 hbase 版本的 Phoenix，將下載文件 apache-phoenix-4.15.0-HBase-1.4-bin.tar 解壓，將 phoenix-4.15.0-HBase-1.4-server.jar 拷貝到 hbase 安裝目錄的 lib 目錄下，注意：每台 regionserver 均需要拷貝，重啟 hbase server 即可，官方如下：
download and expand the latest phoenix-[version]-bin.tar.
- Add the phoenix-[version]-server.jar to the classpath of all HBase region server and master and remove any previous version. An easy way to do this is to copy it into the HBase lib directory (use phoenix-core-[version].jar for Phoenix 3.x)
- restart the region servers
- Add the phoenix-[version]-client.jar to the classpath of any Phoenix client.
- download and setup SQuirrel as your SQL client so you can issue adhoc SQL against your HBase cluster
shell 命令
通過案例，create 表，插入語句，更新語句，刪除語句案例
詳細可參考：https://phoenix.apache.org/faq.html
Phoenix 連接hbase的命令如下，sqlline.py [zookeeper] ：
[node-01@root bin]$ ./sqlline.py node-01
Setting property: [isolation, TRANSACTION_READ_COMMITTED]
issuing: !connect jdbc:phoenix:10.35.66.72 none none org.apache.phoenix.jdbc.PhoenixDriver
Connecting to jdbc:phoenix:node-01
15/06/24 13:06:29 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
Connected to: Phoenix (version 4.15)
Driver: PhoenixEmbeddedDriver (version 4.15)
Autocommit status: true
Transaction isolation: TRANSACTION_READ_COMMITTED
Building list of tables and columns for tab-completion (set fastconnect to true to skip)...
193/193 (100%) Done
Done
sqlline version 1.1.2
0: jdbc:phoenix:node-01>!tables
+------------------------------------------+------------------------------------------+------------------------------------------+-------------------+
| TABLE_CAT | TABLE_SCHEM | TABLE_NAME | TA |
+------------------------------------------+------------------------------------------+------------------------------------------+-------------------+
| null | WL | BIG_LOG_DEVUTRACEID_INDEX | INDEX |
| null | WL | MSGCENTER_PUSHMESSAGE_V2_OWNERPAGE_INDEX | INDEX |
| null | SYSTEM | CATALOG | SYSTEM TABLE |
| null | SYSTEM | SEQUENCE | SYSTEM TABLE |
| null | SYSTEM | STATS | SYSTEM TABLE |
| null | DMO | SOWNTOWN_STATICS | TABLE |
| null | OL | BIGLOG | TABLE |
| null | WL | BIG_LOG | TABLE |
| null | WL | ERROR_LOG | TABLE |
| null | WL | MSGCENTER_PUSHMESSAGE | TABLE |
| null | WL | MSGCENTER_PUSHMESSAGE_V2 | TABLE |
+------------------------------------------+------------------------------------------+------------------------------------------+----------------
從上面能夠看到，已經連接到了hbase集群上面，輸入Phoenix支持的命令 ！tables 可以查看當前集群中存在的數據表，能夠看到有些是 SYSTEM TABLE，其它的都是自己建立的；
下面通過腳本來模擬下使用Phoenix建立數據表、修改表、添加數據、修改數據、刪除數據、刪除表等操作：
建表
新建一張Person表，含有IDCardNum，Name，Age 三個字段 ，test 為 table_schem ，標准sql如下：
create table IF NOT EXISTS test.Person (IDCardNum INTEGER not null primary key, Name varchar(20),Age INTEGER);
在 Phoenix 中使用如下：
0: jdbc:phoenix:10.35.66.72> create table IF NOT EXISTS test.Person (IDCardNum INTEGER not null primary key, Name varchar(20),Age INTEGER);
No rows affected (0.344 seconds)
0: jdbc:phoenix:10.35.66.72> !tables
+------------------------------------------+------------------------------------------+------------------------------------------+-------------------+
| TABLE_CAT | TABLE_SCHEM | TABLE_NAME | TA |
+------------------------------------------+------------------------------------------+------------------------------------------+-------------------+
| null | WL | BIG_LOG_DEVUTRACEID_INDEX | INDEX |
| null | WL | MSGCENTER_PUSHMESSAGE_V2_OWNERPAGE_INDEX | INDEX |
| null | SYSTEM | CATALOG | SYSTEM TABLE |
| null | SYSTEM | SEQUENCE | SYSTEM TABLE |
| null | SYSTEM | STATS | SYSTEM TABLE |
| null | DMO | SOWNTOWN_STATICS | TABLE |
| null | OL | BIGLOG | TABLE |
| null | TEST | PERSON | TABLE |
| null | WL | BIG_LOG | TABLE |
| null | WL | ERROR_LOG | TABLE |
| null | WL | MSGCENTER_PUSHMESSAGE | TABLE |
| null | WL | MSGCENTER_PUSHMESSAGE_V2 | TABLE |
+------------------------------------------+------------------------------------------+------------------------------------------+-------------------+
0: jdbc:phoenix:10.35.66.72> select * from TEST.PERSON;
+------------------------------------------+----------------------+------------------------------------------+
| IDCARDNUM | NAME | AGE |
+------------------------------------------+----------------------+------------------------------------------+
+------------------------------------------+----------------------+------------------------------------------+
可以看到，hbase中已經存在數據表 Person了，包含了三列。
插入數據
對表進行插入操作，sql如下：
insert into Person (IDCardNum,Name,Age) values (100,'小明',12);
insert into Person (IDCardNum,Name,Age) values (101,'小紅',15);
insert into Person (IDCardNum,Name,Age) values (103,'小王',22);
在 Phoenix 中插入的語句為 upsert ，具體如下：
0: jdbc:phoenix:10.35.66.72> upsert into test.Person (IDCardNum,Name,Age) values (100,'小明',12);
1 row affected (0.043 seconds)
0: jdbc:phoenix:10.35.66.72> upsert into test.Person (IDCardNum,Name,Age) values (101,'小紅',15);
1 row affected (0.018 seconds)
0: jdbc:phoenix:10.35.66.72> upsert into test.Person (IDCardNum,Name,Age) values (103,'小王',22);
1 row affected (0.009 seconds)
0: jdbc:phoenix:10.35.66.72> select * from test.Person;
+------------------------------------------+----------------------+------------------------------------------+
| IDCARDNUM | NAME | AGE |
+------------------------------------------+----------------------+------------------------------------------+
| 100 | 小明 | 12 |
| 101 | 小紅 | 15 |
| 103 | 小王 | 22 |
+------------------------------------------+----------------------+------------------------------------------+
3 rows selected (0.115 seconds)
從上面可以看到，三條數據已經進入hbase里面了；好了，現在要對表添加一列 sex 性別操作，怎么辦？
修改數據
alter 修改表數據，sql如下：
ALTER TABLE test.Persion ADD sex varchar(10);
Phoenix 中操作如下：
0: jdbc:phoenix:10.35.66.72> ALTER TABLE test.Person ADD sex varchar(10);
No rows affected (0.191 seconds)
: jdbc:phoenix:10.35.66.72> select * from test.person;
+------------------------------------------+----------------------+------------------------------------------+------------+
| IDCARDNUM | NAME | AGE | SEX |
+------------------------------------------+----------------------+------------------------------------------+------------+
| 100 | 小明 | 12 | null |
| 101 | 小紅 | 15 | null |
| 103 | 小王 | 22 | null |
+------------------------------------------+----------------------+------------------------------------------+------------+
3 rows selected (0.113 seconds)
上圖看到已經新增了列sex，每行的默認值為 null ，那么怎么樣修改這些值呢？
更新表數據 ，標准的sql 如下：
update test.Person set sex='男' where IDCardNum=100;
update test.Person set sex='女' where IDCardNum=101;
update test.Person set sex='男' where IDCardNum=103;
Phoenix中不存在update的語法關鍵字，而是upsert ，功能上替代了Insert+update，官方說明為：
UPSERT VALUES
Inserts if not present and updates otherwise the value in the table. The list of columns is optional and if not present, the values will map to the column in the order they are declared in the schema. The values must evaluate to constants.
根據介紹，只需要在upsert語句中制定存在的idcardnum即可實現更新，在 Phoenix 客戶端中操作如下
0: jdbc:phoenix:10.35.66.72> upsert into test.person (idcardnum,sex) values (100,'男');
1 row affected (0.083 seconds)
0: jdbc:phoenix:10.35.66.72> upsert into test.person (idcardnum,sex) values (101,'女');
1 row affected (0.012 seconds)
0: jdbc:phoenix:10.35.66.72> upsert into test.person (idcardnum,sex) values (103,'男');
1 row affected (0.008 seconds)
0: jdbc:phoenix:10.35.66.72> select * from test.person;
+------------------------------------------+----------------------+------------------------------------------+------------+
| IDCARDNUM | NAME | AGE | SEX |
+------------------------------------------+----------------------+------------------------------------------+------------+
| 100 | 小明 | 12 | 男 |
| 101 | 小紅 | 15 | 女 |
| 103 | 小王 | 22 | 男 |
+------------------------------------------+----------------------+------------------------------------------+------------+
3 rows selected (0.087 seconds)
查詢數據
通過Phoenix可以支持 where、group by、case when 等復雜的查詢條件，案例如下：
# 現增加幾條數據
0: jdbc:phoenix:10.35.66.72> upsert into test.Person (IDCardNum,Name,Age,sex) values (104,'小張',23,'男');
1 row affected (0.012 seconds)
0: jdbc:phoenix:10.35.66.72> upsert into test.Person (IDCardNum,Name,Age,sex) values (105,'小李',28,'男');
1 row affected (0.015 seconds)
0: jdbc:phoenix:10.35.66.72> upsert into test.Person (IDCardNum,Name,Age,sex) values (106,'小李',33,'男');
1 row affected (0.011 seconds)
0: jdbc:phoenix:10.35.66.72> select * from test.person;
+------------------------------------------+----------------------+------------------------------------------+------------+
| IDCARDNUM | NAME | AGE | SEX |
+------------------------------------------+----------------------+------------------------------------------+------------+
| 100 | 小明 | 12 | 男 |
| 101 | 小紅 | 15 | 女 |
| 103 | 小王 | 22 | 男 |
| 104 | 小張 | 23 | 男 |
| 105 | 小李 | 28 | 男 |
| 106 | 小李 | 33 | 男 |
+------------------------------------------+----------------------+------------------------------------------+------------+
6 rows selected (0.09 seconds)
where + group by 語句例子：
jdbc:phoenix:10.35.66.72> select sex ,count(sex) as num from test.person where age >20 group by sex;
+------------+------------------------------------------+
| SEX | NUM |
+------------+------------------------------------------+
| 男 | 4 |
+------------+------------------------------------------+
case when 的例子：
0: jdbc:phoenix:10.35.66.72> select (case name when '小明' then '明明啊' when '小紅' then '紅紅啊' else name end) as showname from test.person;
+------------------------------------------+
| SHOWNAME |
+------------------------------------------+
| 明明啊 |
| 紅紅啊 |
| 小王 |
| 小張 |
| 小李 |
| 小李 |
+------------------------------------------+
更多支持語法參考：https://phoenix.apache.org/language/index.html
刪除數據
刪除數據及刪除表，標准sql如下：
delete from test.Person where idcardnum=100;
drop table test.person;
Phoenix中同標准sql一樣，案例如下：
0: jdbc:phoenix:10.35.66.72> delete from test.Person where idcardnum=100;
1 row affected (0.072 seconds)
0: jdbc:phoenix:10.35.66.72> select * from test.Person where idcardnum=100;
+------------------------------------------+----------------------+------------------------------------------+------------+
| IDCARDNUM | NAME | AGE | SEX |
+------------------------------------------+----------------------+------------------------------------------+------------+
+------------------------------------------+----------------------+------------------------------------------+------------+
0: jdbc:phoenix:10.35.66.72> drop table test.person;
No rows affected (1.799 seconds)
0: jdbc:phoenix:10.35.66.72> select * from test.person;
Error: ERROR 1012 (42M03): Table undefined. tableName=TEST.PERSON (state=42M03,code=1012)
0: jdbc:phoenix:10.35.66.72> !tables
+------------------------------------------+------------------------------------------+------------------------------------------+-------------------+
| TABLE_CAT | TABLE_SCHEM | TABLE_NAME | TA |
+------------------------------------------+------------------------------------------+------------------------------------------+-------------------+
| null | WL | BIG_LOG_DEVUTRACEID_INDEX | INDEX |
| null | WL | MSGCENTER_PUSHMESSAGE_V2_OWNERPAGE_INDEX | INDEX |
| null | SYSTEM | CATALOG | SYSTEM TABLE |
| null | SYSTEM | SEQUENCE | SYSTEM TABLE |
| null | SYSTEM | STATS | SYSTEM TABLE |
| null | DMO | SOWNTOWN_STATICS | TABLE |
| null | OL | BIGLOG | TABLE |
| null | WL | BIG_LOG | TABLE |
| null | WL | ERROR_LOG | TABLE |
| null | WL | MSGCENTER_PUSHMESSAGE | TABLE |
| null | WL | MSGCENTER_PUSHMESSAGE_V2 | TABLE |
+------------------------------------------+------------------------------------------+------------------------------------------+-------------------+
圖形化客戶端SQuirrel使用
如果你不喜歡 終端下的腳本命令，青睞於GUI化的客戶端，那么 SQuirrel是個好的選擇，就跟平日里使用 MsSqlServer client、Navicat client 一樣，效果如下圖：
使用方法：（可以參照官網英文說明）
- 下載SQuirrel 客戶端 ，地址 http://squirrel-sql.sourceforge.net/
- 解壓縮，刪除 lib/ 下老版本的 phoenix-[oldversion]-client.jar 文件，將你剛剛下載的Phoenix文件夾下最新的文件拷貝過去；
- 啟動 SQuirrel 客戶端，選擇 Drivers-new driver ，名稱隨便，url格式： jdbc:phoenix:（zk地址） ，class name textbox 填寫org.apache.phoenix.jdbc.PhoenixDriver
- ok， 點擊 connect即可完成連接
squirrel 客戶端的用法 和 Phoenix 自帶終端一樣，都是常見的 sql 語句，大家可以自己搭建練習。
java client api 使用
創建 maven 項目導入 phoenix 依賴包
<!-- https://mvnrepository.com/artifact/org.apache.phoenix/phoenix-client -->
<dependency>
<groupId>org.apache.phoenix</groupId>
<artifactId>phoenix-client</artifactId>
<version>4.15.0-HBase-1.4</version>
</dependency>
java api 完全可以采用傳統的 jdbc 連接的形式，案例如官方提供：
創建test.java 類，內容如下：
import java.sql.Connection;
import java.sql.DriverManager;
import java.sql.ResultSet;
import java.sql.SQLException;
import java.sql.PreparedStatement;
import java.sql.Statement;
public class test {
public static void main(String[] args) throws SQLException {
Statement stmt = null;
ResultSet rset = null;
Connection con = DriverManager.getConnection("jdbc:phoenix:[zookeeper]");
stmt = con.createStatement();
stmt.executeUpdate("create table test (mykey integer not null primary key, mycolumn varchar)");
stmt.executeUpdate("upsert into test values (1,'Hello')");
stmt.executeUpdate("upsert into test values (2,'World!')");
con.commit();
PreparedStatement statement = con.prepareStatement("select * from test");
rset = statement.executeQuery();
while (rset.next()) {
System.out.println(rset.getString("mycolumn"));
}
statement.close();
con.close();
}
}