习惯了MSSQL（并且可能被宠坏了），我想知道如何获得Oracle 10g 的表大小。
我用谷歌搜索了它，所以我现在知道我可能没有像sp_spaceused 这样简单的选项。我得到的潜在答案大部分时间仍然过时或不起作用。可能是因为我不是我正在使用的架构的 DBA。
有人有解决方案和/或建议吗？
【问题讨论】：
标签： oracle
您可能对此查询感兴趣。它告诉您为每个表分配了多少空间，同时考虑到表上的索引和任何 LOB。您通常有兴趣了解“Purchase Order 表占用多少空间，包括任何索引”，而不仅仅是表本身。你总是可以深入研究细节。请注意，这需要访问 DBA_* 视图。
COLUMN TABLE_NAME FORMAT A32
COLUMN OBJECT_NAME FORMAT A32
COLUMN OWNER FORMAT A10
SELECT
owner,
table_name,
TRUNC(sum(bytes)/1024/1024) Meg,
ROUND( ratio_to_report( sum(bytes) ) over () * 100) Percent
FROM
(SELECT segment_name table_name, owner, bytes
FROM dba_segments
WHERE segment_type IN ('TABLE', 'TABLE PARTITION', 'TABLE SUBPARTITION')
UNION ALL
SELECT i.table_name, i.owner, s.bytes
FROM dba_indexes i, dba_segments s
WHERE s.segment_name = i.index_name
AND s.owner = i.owner
AND s.segment_type IN ('INDEX', 'INDEX PARTITION', 'INDEX SUBPARTITION')
UNION ALL
SELECT l.table_name, l.owner, s.bytes
FROM dba_lobs l, dba_segments s
WHERE s.segment_name = l.segment_name
AND s.owner = l.owner
AND s.segment_type IN ('LOBSEGMENT', 'LOB PARTITION')
UNION ALL
SELECT l.table_name, l.owner, s.bytes
FROM dba_lobs l, dba_segments s
WHERE s.segment_name = l.index_name
AND s.owner = l.owner
AND s.segment_type = 'LOBINDEX')
WHERE owner in UPPER('&owner')
GROUP BY table_name, owner
HAVING SUM(bytes)/1024/1024 > 10 /* Ignore really small tables */
ORDER BY SUM(bytes) desc
;
【讨论】：
-- Tables + Size MB
select owner, table_name, round((num_rows*avg_row_len)/(1024*1024)) MB
from all_tables
where owner not like 'SYS%' -- Exclude system tables.
and num_rows > 0 -- Ignore empty Tables.
order by MB desc -- Biggest first.
;
--Tables + Rows
select owner, table_name, num_rows
from all_tables
where owner not like 'SYS%' -- Exclude system tables.
and num_rows > 0 -- Ignore empty Tables.
order by num_rows desc -- Biggest first.
;
注意：这些是估计值，通过收集统计数据更加准确：
exec dbms_utility.analyze_schema(user,'COMPUTE');
【讨论】：
null(num_rows,avg_row_len)，需要先通过下面的语句ANALYZE TABLE your_table COMPUTE STATISTICS做一些分析
首先，我通常会警告说，为了进行空间分析而收集表统计信息是一件有潜在危险的事情。收集统计信息可能会更改查询计划，特别是如果 DBA 配置了一个使用您的调用未使用的非默认参数的统计信息收集作业，并且会导致 Oracle 重新解析使用相关表的查询，这可能是一种性能打。如果 DBA 故意留下一些没有统计信息的表（如果您的 OPTIMIZER_MODE 是 CHOOSE 则很常见），收集统计信息可能会导致 Oracle 停止使用基于规则的优化器并开始使用基于成本的优化器进行一组查询，这可能是如果在生产中意外完成，则会导致严重的性能问题。如果您的统计准确，您可以直接查询USER_TABLES（或ALL_TABLES或DBA_TABLES），无需调用GATHER_TABLE_STATS。如果您的统计数据不准确，可能是有原因的，您不想破坏现状。
其次，与 SQL Server sp_spaceused 过程最接近的等价物可能是 Oracle 的 DBMS_SPACE 包。 Tom Kyte 有一个不错的show_space procedure，它为这个包提供了一个简单的界面，并打印出类似于sp_spaceused 打印出来的信息。
【讨论】：
首先，收集表格上的优化器统计信息（如果您还没有）：
begin
dbms_stats.gather_table_stats('MYSCHEMA','MYTABLE');
end;
/
警告：正如贾斯汀在他的回答中所说，收集优化器统计信息会影响查询优化，不应在没有适当注意和考虑的情况下进行！
然后从生成的stats中找出该表占用的块数：
select blocks, empty_blocks, num_freelist_blocks
from all_tables
where owner = 'MYSCHEMA'
and table_name = 'MYTABLE';
分配给表的总块数为blocks + empty_blocks + num_freelist_blocks。
blocks 是实际包含数据的块数。
将块数乘以正在使用的块大小（通常为 8KB）以获得消耗的空间 - 例如17 个块 x 8KB = 136KB。
要同时对架构中的所有表执行此操作：
begin
dbms_stats.gather_schema_stats ('MYSCHEMA');
end;
/
select table_name, blocks, empty_blocks, num_freelist_blocks
from user_tables;
注意：阅读this AskTom thread后对上述内容的更改
【讨论】：
我修改了 WW 的查询以提供更详细的信息：
SELECT * FROM (
SELECT
owner, object_name, object_type, table_name, ROUND(bytes)/1024/1024 AS meg,
tablespace_name, extents, initial_extent,
ROUND(Sum(bytes/1024/1024) OVER (PARTITION BY table_name)) AS total_table_meg
FROM (
-- Tables
SELECT owner, segment_name AS object_name, 'TABLE' AS object_type,
segment_name AS table_name, bytes,
tablespace_name, extents, initial_extent
FROM dba_segments
WHERE segment_type IN ('TABLE', 'TABLE PARTITION', 'TABLE SUBPARTITION')
UNION ALL
-- Indexes
SELECT i.owner, i.index_name AS object_name, 'INDEX' AS object_type,
i.table_name, s.bytes,
s.tablespace_name, s.extents, s.initial_extent
FROM dba_indexes i, dba_segments s
WHERE s.segment_name = i.index_name
AND s.owner = i.owner
AND s.segment_type IN ('INDEX', 'INDEX PARTITION', 'INDEX SUBPARTITION')
-- LOB Segments
UNION ALL
SELECT l.owner, l.column_name AS object_name, 'LOB_COLUMN' AS object_type,
l.table_name, s.bytes,
s.tablespace_name, s.extents, s.initial_extent
FROM dba_lobs l, dba_segments s
WHERE s.segment_name = l.segment_name
AND s.owner = l.owner
AND s.segment_type = 'LOBSEGMENT'
-- LOB Indexes
UNION ALL
SELECT l.owner, l.column_name AS object_name, 'LOB_INDEX' AS object_type,
l.table_name, s.bytes,
s.tablespace_name, s.extents, s.initial_extent
FROM dba_lobs l, dba_segments s
WHERE s.segment_name = l.index_name
AND s.owner = l.owner
AND s.segment_type = 'LOBINDEX'
)
WHERE owner = UPPER('&owner')
)
WHERE total_table_meg > 10
ORDER BY total_table_meg DESC, meg DESC
/
【讨论】：
IIRC 您需要的表是 DBA_TABLES、DBA_EXTENTS 或 DBA_SEGMENTS 和 DBA_DATA_FILES。如果您没有计算机上的管理权限，您可以查看这些表的 USER_ 和 ALL_ 版本。
【讨论】：
对于子分区表和索引，我们可以使用以下查询
SELECT owner, table_name, ROUND(sum(bytes)/1024/1024/1024, 2) GB
FROM
(SELECT segment_name table_name, owner, bytes
FROM dba_segments
WHERE segment_type IN ('TABLE', 'TABLE PARTITION', 'TABLE SUBPARTITION')
UNION ALL
SELECT i.table_name, i.owner, s.bytes
FROM dba_indexes i, dba_segments s
WHERE s.segment_name = i.index_name
AND s.owner = i.owner
AND s.segment_type IN ('INDEX', 'INDEX PARTITION', 'INDEX SUBPARTITION')
UNION ALL
SELECT l.table_name, l.owner, s.bytes
FROM dba_lobs l, dba_segments s
WHERE s.segment_name = l.segment_name
AND s.owner = l.owner
AND s.segment_type = 'LOBSEGMENT'
UNION ALL
SELECT l.table_name, l.owner, s.bytes
FROM dba_lobs l, dba_segments s
WHERE s.segment_name = l.index_name
AND s.owner = l.owner
AND s.segment_type = 'LOBINDEX')
WHERE owner in UPPER('&owner')
GROUP BY table_name, owner
HAVING SUM(bytes)/1024/1024 > 10 /* Ignore really small tables */
ORDER BY SUM(bytes) DESC
;
【讨论】：
select segment_name,segment_type,bytes/1024/1024 MB
from dba_segments
where segment_name='TABLENAME' and owner ='OWNERNAME' order by mb desc;
【讨论】：
这是 WWs 答案的变体，它包括上面其他人建议的分区和子分区，以及显示类型的列：表/索引/LOB 等
SELECT
owner, "Type", table_name "Name", TRUNC(sum(bytes)/1024/1024) Meg
FROM
( SELECT segment_name table_name, owner, bytes, 'Table' as "Type"
FROM dba_segments
WHERE segment_type in ('TABLE','TABLE PARTITION','TABLE SUBPARTITION')
UNION ALL
SELECT i.table_name, i.owner, s.bytes, 'Index' as "Type"
FROM dba_indexes i, dba_segments s
WHERE s.segment_name = i.index_name
AND s.owner = i.owner
AND s.segment_type in ('INDEX','INDEX PARTITION','INDEX SUBPARTITION')
UNION ALL
SELECT l.table_name, l.owner, s.bytes, 'LOB' as "Type"
FROM dba_lobs l, dba_segments s
WHERE s.segment_name = l.segment_name
AND s.owner = l.owner
AND s.segment_type IN ('LOBSEGMENT','LOB PARTITION','LOB SUBPARTITION')
UNION ALL
SELECT l.table_name, l.owner, s.bytes, 'LOB Index' as "Type"
FROM dba_lobs l, dba_segments s
WHERE s.segment_name = l.index_name
AND s.owner = l.owner
AND s.segment_type = 'LOBINDEX')
WHERE owner in UPPER('&owner')
GROUP BY table_name, owner, "Type"
HAVING SUM(bytes)/1024/1024 > 10 /* Ignore really small tables */
ORDER BY SUM(bytes) desc;
【讨论】：
我修改了查询以获取每个表空间的架构大小..
SELECT owner,
tablespace_name,
TRUNC (SUM (bytes) / 1024 / 1024) Meg,
ROUND (ratio_to_report (SUM (bytes)) OVER () * 100) Percent
FROM (SELECT tablespace_name, owner, bytes
FROM dba_segments
WHERE segment_type IN
('TABLE', 'TABLE PARTITION', 'TABLE SUBPARTITION')
UNION ALL
SELECT i.tablespace_name, i.owner, s.bytes
FROM dba_indexes i, dba_segments s
WHERE s.segment_name = i.index_name
AND s.owner = i.owner
AND s.segment_type IN
('INDEX', 'INDEX PARTITION', 'INDEX SUBPARTITION')
UNION ALL
SELECT l.tablespace_name, l.owner, s.bytes
FROM dba_lobs l, dba_segments s
WHERE s.segment_name = l.segment_name
AND s.owner = l.owner
AND s.segment_type IN ('LOBSEGMENT', 'LOB PARTITION')
UNION ALL
SELECT l.tablespace_name, l.owner, s.bytes
FROM dba_lobs l, dba_segments s
WHERE s.segment_name = l.index_name
AND s.owner = l.owner
AND s.segment_type = 'LOBINDEX')
WHERE owner IN UPPER ('&owner')
GROUP BY owner, tablespace_name
--HAVING SUM(bytes)/1024/1024 > 10 /* Ignore really small tables */
ORDER BY tablespace_name -- desc
;
【讨论】：
取决于您所说的“表格大小”。 表与文件系统上的特定文件无关。一个表将驻留在一个表空间上（如果它是分区的，则可能是多个表空间，如果您还想考虑表上的索引，则可能是多个表空间）。 一个表空间中通常会有多个表，并且可能分布在多个文件中。
如果您要估计表的未来增长需要多少空间，那么 avg_row_len 乘以表中的行数（或您期望的表中的行数）将是一个很好的指南。 但是 Oracle 会在每个块上留出一些空闲空间，部分是为了允许行在更新时“增长”，部分是因为在该块上可能无法容纳另一整行（例如，8K 块只能容纳 2 行3K，尽管这将是一个极端的例子，因为 3K 比大多数行大小大得多）。 所以 BLOCKS（在 USER_TABLES 中）可能是一个更好的指南。
但是，如果您在一个表中有 200,000 行，删除了其中的一半，那么该表仍将“拥有”相同数量的块。它不会释放它们以供其他表使用。 此外，块不会单独添加到表中，而是以称为“范围”的组的形式添加。因此，表中通常会有 EMPTY_BLOCKS（也在 USER_TABLES 中）。
【讨论】：
更正分区表：
SELECT owner, table_name, ROUND(sum(bytes)/1024/1024/1024, 2) GB FROM (SELECT segment_name table_name, owner, bytes FROM dba_segments WHERE segment_type IN ('TABLE', 'TABLE PARTITION', 'TABLE SUBPARTITION') UNION ALL SELECT i.table_name, i.owner, s.bytes FROM dba_indexes i, dba_segments s WHERE s.segment_name = i.index_name AND s.owner = i.owner AND s.segment_type IN ('INDEX', 'INDEX PARTITION', 'INDEX SUBPARTITION') UNION ALL SELECT l.table_name, l.owner, s.bytes FROM dba_lobs l, dba_segments s WHERE s.segment_name = l.segment_name and s.owner = l.owner AND s.segment_type in ('LOBSEGMENT', 'LOB PARTITION', 'LOB SUBPARTITION') UNION ALL SELECT l.table_name, l.owner, s.bytes FROM dba_lobs l, dba_segments s WHERE s.segment_name = l.index_name AND s.owner = l.owner AND s.segment_type = 'LOBINDEX') WHERE owner in UPPER('&owner') GROUP BY table_name, owner HAVING SUM(bytes)/1024/1024 > 10 /* Ignore really small tables */ order by sum(bytes) desc ;
【讨论】：
根据块大小返回表的原始大小的简单选择，还包括带索引的大小
选择表名,(nvl (( 选择总和（块） 从 dba_indexes a,dba_segments b 其中 a.index_name=b.segment_name 和 a.table_name=dba_tables.table_name ),0)+blocks)*8192/1024 TotalSize,blocks*8 tableSize 来自 dba_tables 3 前下单
【讨论】：
我发现这更准确一点：
SELECT
owner, table_name, TRUNC(sum(bytes)/1024/1024/1024) GB
FROM
(SELECT segment_name table_name, owner, bytes
FROM dba_segments
WHERE segment_type in ('TABLE','TABLE PARTITION')
UNION ALL
SELECT i.table_name, i.owner, s.bytes
FROM dba_indexes i, dba_segments s
WHERE s.segment_name = i.index_name
AND s.owner = i.owner
AND s.segment_type in ('INDEX','INDEX PARTITION')
UNION ALL
SELECT l.table_name, l.owner, s.bytes
FROM dba_lobs l, dba_segments s
WHERE s.segment_name = l.segment_name
AND s.owner = l.owner
AND s.segment_type IN ('LOBSEGMENT','LOB PARTITION')
UNION ALL
SELECT l.table_name, l.owner, s.bytes
FROM dba_lobs l, dba_segments s
WHERE s.segment_name = l.index_name
AND s.owner = l.owner
AND s.segment_type = 'LOBINDEX')
---WHERE owner in UPPER('&owner')
GROUP BY table_name, owner
HAVING SUM(bytes)/1024/1024 > 10 /* Ignore really small tables */
ORDER BY SUM(bytes) desc
【讨论】：
select segment_name as tablename, sum(bytes/ (1024 * 1024 * 1024)) as tablesize_in_GB
From dba_segments /* if looking at tables not owned by you else use user_segments */
where segment_name = 'TABLE_WHOSE_SIZE_I_WANT_TO_KNOW'
and OWNER = 'WHO OWNS THAT TABLE' /* if user_segments is used delete this line */
group by segment_name ;
【讨论】：
还有一个选项允许通过连接获得“选择”大小，以及表格大小作为选项
-- 1
EXPLAIN PLAN
FOR
SELECT
Scheme.Table_name.table_column1 AS "column1",
Scheme.Table_name.table_column2 AS "column2",
Scheme.Table_name.table_column3 AS "column3",
FROM Scheme.Table_name
WHERE ;
SELECT * FROM TABLE (DBMS_XPLAN.display);
【讨论】：
我有与最后一个计算表数据段、表索引和 blob 字段的变体相同的变体：
CREATE OR REPLACE FUNCTION
SYS.RAZMER_TABLICY_RAW(pNazvanie in varchar, pOwner in varchar2)
return number
is
val number(16);
sz number(16);
begin
sz := 0;
--Calculate size of table data segments
select
sum(t.bytes) into val
from
sys.dba_segments t
where
t.segment_name = upper(pNazvanie)
and
t.owner = upper(pOwner);
sz := sz + nvl(val,0);
--Calculate size of table indexes segments
select
sum(s.bytes) into val
from
all_indexes t
inner join
dba_segments s
on
t.index_name = s.segment_name
where
t.table_name = upper(pNazvanie)
and
t.owner = upper(pOwner);
sz := sz + nvl(val,0);
--Calculate size of table blob segments
select
sum(s.bytes) into val
from
all_lobs t
inner join
dba_segments s on t.segment_name = s.segment_name
where
t.table_name = upper(pNazvanie)
and
t.owner = upper(pOwner);
sz := sz + nvl(val,0);
return sz;
end razmer_tablicy_raw;
【讨论】：