溫馨提示×
您好，登錄后才能下訂單哦！
點擊 登錄注冊 即表示同意《億速云用戶服務條款》
您好，登錄后才能下訂單哦！
如何理解ADD和DROP分區(qū)，針對這個問題，這篇文章詳細介紹了相對應的分析和解答，希望可以幫助更多想解決這個問題的小伙伴找到更簡單易行的方法。
add和drop分區(qū)
語法：
ALTER TABLE t_pe_r
ADD PARTITION p4 VALUES LESS THAN (30 );
ALTER TABLE t_pe_l
ADD PARTITION p4 VALUES (30 );
ALTER TABLE t_pe_h
ADD PARTITION p3;
alter table t_pe_r drop partition p3;
限制：如果范圍分區(qū)使用maxvalue選項則報錯ORA-14074: 分區(qū)界限必須調(diào)整為高于最后一個分區(qū)界限
如果LIST分區(qū)使用了default選項則報錯ORA-14323: 在 DEFAULT 分區(qū)已存在時無法添加分區(qū)
drop 分區(qū)不能使用在HASH分區(qū)表中，報錯ORA-14255: 未按范圍, 組合范圍或列表方法對表進行分區(qū)
我們這里討論HASH、list、range 3方式下add partition和drop partition關于local索引，global索引和普通索引的狀態(tài)。
使用腳本
drop table t_pe_r ;
drop table t_pe_l;
drop table t_pe_h;
CREATE TABLE t_pe_r (i NUMBER, j NUMBER , f varchar2(20),k varchar2(20))
PARTITION BY RANGE(j)
(PARTITION p1 VALUES LESS THAN (10),
PARTITION p2 VALUES LESS THAN (20));
create index t_pe_r_n on t_pe_r(i);
create index t_pe_r_l on t_pe_r(j) local;
create index t_pe_r_g on t_pe_r(f)
GLOBAL PARTITION BY hash (f)
(partition pg1 ,
partition pg2);
CREATE TABLE t_pe_l (i NUMBER, j NUMBER , f varchar2(20),k varchar2(20))
PARTITION BY list (j)
(PARTITION p1 VALUES (10),
PARTITION p2 VALUES (20));
create index t_pe_l_n on t_pe_l(i);
create index t_pe_l_l on t_pe_l(j) local;
create index t_pe_l_g on t_pe_l(f)
GLOBAL PARTITION BY hash (f)
(partition pg1 ,
partition pg2);
CREATE TABLE t_pe_h (i NUMBER, j NUMBER , f varchar2(20),k varchar2(20))
PARTITION BY hash(j)
(PARTITION p1 ,
PARTITION p2 );
create index t_pe_h_n on t_pe_h(i);
create index t_pe_h_l on t_pe_h(j) local;
create index t_pe_h_g on t_pe_h(f)
GLOBAL PARTITION BY hash (f)
(partition pg1 ,
partition pg2);
insert into t_pe_r
values(2,5,'a','A');
insert into t_pe_r
values(1,15,'b','B');
insert into t_pe_l
values(2,10,'a','A');
insert into t_pe_l
values(1,20,'b','B');
insert into t_pe_h
values(2,5,'a','A');
insert into t_pe_h
values(1,15,'b','B');
insert into t_pe_h
values(3,25,'c','C');
select INDEX_NAME ,STATUS,a.partition_name from dba_ind_partitions a where lower(index_name)='t_pe_r_l';
select INDEX_NAME ,STATUS,a.partition_name from dba_ind_partitions a where lower(index_name)='t_pe_r_g';
select INDEX_NAME ,STATUS from dba_indexes where lower(index_name)='t_pe_r_n';
select INDEX_NAME ,STATUS,a.partition_name from dba_ind_partitions a where lower(index_name)='t_pe_l_l';
select INDEX_NAME ,STATUS,a.partition_name from dba_ind_partitions a where lower(index_name)='t_pe_l_g';
select INDEX_NAME ,STATUS from dba_indexes where lower(index_name)='t_pe_l_n';
select INDEX_NAME ,STATUS,a.partition_name from dba_ind_partitions a where lower(index_name)='t_pe_h_l';
select INDEX_NAME ,STATUS,a.partition_name from dba_ind_partitions a where lower(index_name)='t_pe_h_g';
select INDEX_NAME ,STATUS from dba_indexes where lower(index_name)='t_pe_h_n';
下面進行添加
ALTER TABLE t_pe_r
ADD PARTITION p4 VALUES LESS THAN (30 );
ALTER TABLE t_pe_l
ADD PARTITION p4 VALUES (30 );
ALTER TABLE t_pe_h
ADD PARTITION p3;
然后查看索引狀態(tài)
SQL> select INDEX_NAME ,STATUS,a.partition_name from dba_ind_partitions a where lower(index_name)='t_pe_r_l';
INDEX_NAME STATUS PARTITION_NAME
------------------------------ -------- ------------------------------
T_PE_R_L USABLE P2
T_PE_R_L USABLE P1
T_PE_R_L USABLE P4
SQL> select INDEX_NAME ,STATUS,a.partition_name from dba_ind_partitions a where lower(index_name)='t_pe_r_g';
INDEX_NAME STATUS PARTITION_NAME
------------------------------ -------- ------------------------------
T_PE_R_G USABLE PG2
T_PE_R_G USABLE PG1
SQL> select INDEX_NAME ,STATUS from dba_indexes where lower(index_name)='t_pe_r_n';
INDEX_NAME STATUS
------------------------------ --------
T_PE_R_N VALID
SQL> select INDEX_NAME ,STATUS,a.partition_name from dba_ind_partitions a where lower(index_name)='t_pe_l_l';
INDEX_NAME STATUS PARTITION_NAME
------------------------------ -------- ------------------------------
T_PE_L_L USABLE P2
T_PE_L_L USABLE P1
T_PE_L_L USABLE P4
SQL> select INDEX_NAME ,STATUS,a.partition_name from dba_ind_partitions a where lower(index_name)='t_pe_l_g';
INDEX_NAME STATUS PARTITION_NAME
------------------------------ -------- ------------------------------
T_PE_L_G USABLE PG2
T_PE_L_G USABLE PG1
SQL> select INDEX_NAME ,STATUS from dba_indexes where lower(index_name)='t_pe_l_n';
INDEX_NAME STATUS
------------------------------ --------
T_PE_L_N VALID
SQL> select INDEX_NAME ,STATUS,a.partition_name from dba_ind_partitions a where lower(index_name)='t_pe_h_l';
INDEX_NAME STATUS PARTITION_NAME
------------------------------ -------- ------------------------------
T_PE_H_L USABLE P2
T_PE_H_L UNUSABLE P1
T_PE_H_L UNUSABLE P3
SQL> select INDEX_NAME ,STATUS,a.partition_name from dba_ind_partitions a where lower(index_name)='t_pe_h_g';
INDEX_NAME STATUS PARTITION_NAME
------------------------------ -------- ------------------------------
T_PE_H_G UNUSABLE PG2
T_PE_H_G UNUSABLE PG1
SQL> select INDEX_NAME ,STATUS from dba_indexes where lower(index_name)='t_pe_h_n';
INDEX_NAME STATUS
------------------------------ --------
T_PE_H_N UNUSABLE
可以看到實際上LIST和HASH的所有索引都沒有受到影響，而HASH分區(qū)則不同，所有的索引均失效，添加分區(qū)后通過HASH算法重新分布了行，那么應該ROWID也受到了影響，可以DUMP出來看看。
索引進行rebuild
alter index T_PE_H_L rebuild partition p1;
在進行HASH分區(qū)的加入分區(qū)時候最好如下：
ALTER TABLE t_pe_h
ADD PARTITION p3 update indexes; 加上UPDATE INDEXES，同時實際上HASH的分區(qū)個數(shù)應該是2的N次方，不然會分布不均勻。
然后我們測試下DROP partition，drop partition只能用于RANGE 和LIST分區(qū)方式，HASH分區(qū)不能使用：
You can drop partitions from range, list, or composite range-list partitioned tables. For hash-partitioned tables, or hash subpartitions of range-hash partitioned
tables, you must perform. a coalesce operation instead.
使用腳本：
alter table t_pe_r drop partition p1;
alter table t_pe_l drop partition p1;
然后觀察：
SQL> select INDEX_NAME ,STATUS,a.partition_name from dba_ind_partitions a where lower(index_name)='t_pe_r_l';
INDEX_NAME STATUS PARTITION_NAME
------------------------------ -------- ------------------------------
T_PE_R_L USABLE P2
T_PE_R_L USABLE P4
SQL> select INDEX_NAME ,STATUS,a.partition_name from dba_ind_partitions a where lower(index_name)='t_pe_r_g';
INDEX_NAME STATUS PARTITION_NAME
------------------------------ -------- ------------------------------
T_PE_R_G UNUSABLE PG2
T_PE_R_G UNUSABLE PG1
SQL> select INDEX_NAME ,STATUS from dba_indexes where lower(index_name)='t_pe_r_n';
INDEX_NAME STATUS
------------------------------ --------
T_PE_R_N UNUSABLE
SQL> select INDEX_NAME ,STATUS,a.partition_name from dba_ind_partitions a where lower(index_name)='t_pe_l_l';
INDEX_NAME STATUS PARTITION_NAME
------------------------------ -------- ------------------------------
T_PE_L_L USABLE P2
T_PE_L_L USABLE P4
SQL> select INDEX_NAME ,STATUS,a.partition_name from dba_ind_partitions a where lower(index_name)='t_pe_l_g';
INDEX_NAME STATUS PARTITION_NAME
------------------------------ -------- ------------------------------
T_PE_L_G UNUSABLE PG2
T_PE_L_G UNUSABLE PG1
SQL> select INDEX_NAME ,STATUS from dba_indexes where lower(index_name)='t_pe_l_n';
INDEX_NAME STATUS
------------------------------ --------
T_PE_L_N UNUSABLE
可以看到普通索引，全局索引均已經(jīng)失效，但是本地索引卻不受影響。
如果我們帶上UPDATE INDEXES會怎么樣？
SQL> alter table t_pe_r drop partition p1 update indexes;
Table altered
SQL> alter table t_pe_l drop partition p1 update indexes;
Table altered
SQL>
SQL> select INDEX_NAME ,STATUS,a.partition_name from dba_ind_partitions a where lower(index_name)='t_pe_r_l';
INDEX_NAME STATUS PARTITION_NAME
------------------------------ -------- ------------------------------
T_PE_R_L USABLE P2
T_PE_R_L USABLE P4
SQL> select INDEX_NAME ,STATUS,a.partition_name from dba_ind_partitions a where lower(index_name)='t_pe_r_g';
INDEX_NAME STATUS PARTITION_NAME
------------------------------ -------- ------------------------------
T_PE_R_G USABLE PG2
T_PE_R_G USABLE PG1
SQL> select INDEX_NAME ,STATUS from dba_indexes where lower(index_name)='t_pe_r_n';
INDEX_NAME STATUS
------------------------------ --------
T_PE_R_N VALID
SQL> select INDEX_NAME ,STATUS,a.partition_name from dba_ind_partitions a where lower(index_name)='t_pe_l_l';
INDEX_NAME STATUS PARTITION_NAME
------------------------------ -------- ------------------------------
T_PE_L_L USABLE P2
T_PE_L_L USABLE P4
SQL> select INDEX_NAME ,STATUS,a.partition_name from dba_ind_partitions a where lower(index_name)='t_pe_l_g';
INDEX_NAME STATUS PARTITION_NAME
------------------------------ -------- ------------------------------
T_PE_L_G USABLE PG2
T_PE_L_G USABLE PG1
SQL> select INDEX_NAME ,STATUS from dba_indexes where lower(index_name)='t_pe_l_n';
INDEX_NAME STATUS
------------------------------ --------
T_PE_L_N VALID
可以看到加上UPDATE INDEXES 就會自動重建失效的索引。
結(jié)論：
1、如果范圍分區(qū)使用maxvalue選項則報錯ORA-14074: 分區(qū)界限必須調(diào)整為高于最后一個分區(qū)界限
2、如果LIST分區(qū)使用了default選項則報錯ORA-14323: 在 DEFAULT 分區(qū)已存在時無法添加分區(qū)
3、drop 分區(qū)不能使用在HASH分區(qū)表中，報錯ORA-14255: 未按范圍, 組合范圍或列表方法對表進行分區(qū)，如果要減少一個HASH分區(qū)表中的分區(qū)需要用ALTER TABLE ... COALESCE PARTITION
4、HASH分區(qū)進行ADD PARTITION操作，普通索引，本地索引，全局索引都會失效，除非使用UPDATE INDEXES，但是LIST、RANGE分區(qū)不受影響
5、LIST,RANGE分區(qū)進行DROP PARTITION操作全局索引及普通索引會失效，但是LOCAL索引不受影響。除非使用UPDATE INDEXES.
6、如果想要為全局索引增加分區(qū)，那這個操作只能對HASH分區(qū)的全局有效，ORA-14640: 添加/合并索引分區(qū)操作只對散列分區(qū)的全局索引有效，但是DROP全局索引的分區(qū)對HASH\LIST\RANGE均有效。
關于如何理解ADD和DROP分區(qū)問題的解答就分享到這里了，希望以上內(nèi)容可以對大家有一定的幫助，如果你還有很多疑惑沒有解開，可以關注億速云行業(yè)資訊頻道了解更多相關知識。
免責聲明：本站發(fā)布的內(nèi)容（圖片、視頻和文字）以原創(chuàng)、轉(zhuǎn)載和分享為主，文章觀點不代表本網(wǎng)站立場，如果涉及侵權請聯(lián)系站長郵箱：is@yisu.com進行舉報，并提供相關證據(jù)，一經(jīng)查實，將立刻刪除涉嫌侵權內(nèi)容。
