【发布时间】：2014-03-05 04:15:48
【问题描述】：
我正在尝试优化以下 teradata 查询。任何人都可以请这个。检索记录需要花费大量时间。
select top 100 I.item_sku_nbr,L.loc_nbr,MIS.MVNDR_PRTY_ID from
QA_US_MASTER_VIEWS.item I,
qa4_US_MASTER_VIEWS.location L,
qa4_US_MASTER_VIEWS.item_str IST,
qa4_US_MASTER_VIEWS.mvndr_item_str MIS
where MIS.str_LOC_ID = L.loc_id and
mis.str_loc_id = IST.str_loc_id and
IST.str_loc_id = L.loc_id and
MIS.ITEM_STAT_CD = IST.ITEM_STAT_CD and
IST.ITEM_ID = I.ITEM_ID and
MIS.ITEM_ID = IST.ITEM_ID and
I.ITEM_STAT_CD = 100 and
IST.curr_rmeth_cd = 2 and
MIS.curr_dsvc_typ_cd = 3 and
MIS.OK_TO_ORD_FLG = 'Y' and
MIS.EFF_END_DT = DATE '9999-12-31' and
IST.EFF_END_DT = DATE '9999-12-31' and
MIS.ACTV_FLG ='Y' and
IST.ACTV_FLG ='Y' and I.ACTV_FLG='Y'
解释QA_US_MASTER.LOCATION的计划：
1) 首先，我们将 QA_US_MASTER.LOCATION 锁定在视图中 qa4_US_MASTER_VIEWS. 访问位置。
2) 接下来，我们从 QA_US_MASTER.LOCATION 执行全 AMP 检索步骤 通过全行扫描查看 qa4_US_MASTER_VIEWS.Location Spool 1 (group_amps) 中没有剩余条件，即 在 AMP 上本地构建。 Spool 1 的大小估计为 高置信度为 10,903 行（1,613,644 字节）。这 此步骤的估计时间为 0.01 秒。
3) 最后，我们向所有涉及的 AMP 发送一个 END TRANSACTION 步骤 正在处理请求。
-> Spool 1 的内容作为以下结果发送回用户 声明 1. 总估计时间为 0.01 秒。
解释 qa4_US_MASTER_VIEWS.item_str 的计划：
1) 首先，我们将 QA_US_MASTER.item_str 锁定在视图中 qa4_US_MASTER_VIEWS.item_str 用于访问。
2) 接下来，我们从 QA_US_MASTER.item_str 执行全 AMP 检索步骤 通过全行扫描查看 qa4_US_MASTER_VIEWS.item_str Spool 1 (group_amps) 中没有剩余条件，即 在 AMP 上本地构建。输入表不会被缓存在 内存，但它有资格进行同步扫描。结果 spool 文件不会缓存在内存中。线轴 1 的尺寸为 低置信度估计为 1,229,047,917 行（ 325,697,698,005 字节）。此步骤的估计时间为 4 分 51 秒。
3) 最后，我们向所有涉及的 AMP 发送一个 END TRANSACTION 步骤 正在处理请求。
-> Spool 1 的内容作为以下结果发送回用户 声明 1. 总估计时间为 4 分 51 秒。
解释QA_US_MASTER.ITEM的计划：
1) 首先，我们在视图 qa4_US_MASTER_VIEWS.item 中锁定 QA_US_MASTER.ITEM 访问。
2) 接下来，我们从 QA_US_MASTER.ITEM 中执行全 AMP 检索步骤 通过全行扫描查看 qa4_US_MASTER_VIEWS.item 剩余条件进入阀芯 1 (group_amps)，它是内置的 本地在 AMP 上。 Spool 1 的尺寸估计为高 置信度为 1,413,284 行（357,560,852 字节）。这 此步骤的估计时间为 0.40 秒。
3) 最后，我们向所有涉及的 AMP 发送一个 END TRANSACTION 步骤 正在处理请求。
-> Spool 1 的内容作为以下结果发送回用户 声明 1. 总估计时间为 0.40 秒。
解释QA_US_MASTER.MVNDR_ITEM_STR 的计划：
1) 首先，我们将 QA_US_MASTER.MVNDR_ITEM_STR 锁定在视图中 qa4_US_MASTER_VIEWS.mvndr_item_str 用于访问。
2) 接下来，我们从 QA_US_MASTER.MVNDR_ITEM_STR 在视图中 qa4_US_MASTER_VIEWS.mvndr_item_str 通过全行扫描 Spool 1 (group_amps) 中没有剩余条件，这是内置的 本地在 AMP 上。输入表不会缓存在内存中， 但它有资格进行同步扫描。结果卷轴 文件不会缓存在内存中。线轴 1 的尺寸为 以高置信度估计为 1,316,279,746 行（ 327,753,656,754 字节）。此步骤的估计时间为 6 分 4 秒。
3) 最后，我们向所有涉及的 AMP 发送一个 END TRANSACTION 步骤 正在处理请求。
-> Spool 1 的内容作为以下结果发送回用户 声明 1. 总估计时间为 6 分 4 秒。
解释整个查询的计划：
1) First, we lock QA_US_MASTER.ITEM in view QA_US_MASTER_VIEWS.item
for access, we lock QA_US_MASTER.LOCATION in view
qa4_US_MASTER_VIEWS.location for access, we lock
QA_US_MASTER.MVNDR_ITEM_STR in view
qa4_US_MASTER_VIEWS.mvndr_item_str for access, and we lock
QA_US_MASTER.item_str in view qa4_US_MASTER_VIEWS.item_str for
access.
2) Next, we execute the following steps in parallel.
1) We do an all-AMPs RETRIEVE step from QA_US_MASTER.LOCATION in
view qa4_US_MASTER_VIEWS.location by way of an all-rows scan
with no residual conditions into Spool 3 (all_amps)
(compressed columns allowed), which is duplicated on all AMPs.
The size of Spool 3 is estimated with high confidence to be
1,013,979 rows (20,279,580 bytes). The estimated time for
this step is 0.03 seconds.
2) We do an all-AMPs RETRIEVE step from QA_US_MASTER.ITEM in
view QA_US_MASTER_VIEWS.item by way of an all-rows scan with
a condition of ("(QA_US_MASTER.ITEM in view
QA_US_MASTER_VIEWS.item.ITEM_STAT_CD = 100) AND
(QA_US_MASTER.ITEM in view QA_US_MASTER_VIEWS.item.ACTV_FLG =
'Y')") into Spool 4 (all_amps) (compressed columns allowed)
fanned out into 14 hash join partitions, which is duplicated
on all AMPs. The size of Spool 4 is estimated with low
confidence to be 30,819,363 rows (678,025,986 bytes). The
estimated time for this step is 0.81 seconds.
3) We do an all-AMPs JOIN step from Spool 3 (Last Use) by way of an
all-rows scan, which is joined to QA_US_MASTER.item_str in view
qa4_US_MASTER_VIEWS.item_str by way of an all-rows scan with a
condition of
("(QA_US_MASTER.item_str in view
qa4_US_MASTER_VIEWS.item_str.CURR_RMETH_CD = 2) AND
((QA_US_MASTER.item_str in view
qa4_US_MASTER_VIEWS.item_str.EFF_END_DT = DATE '9999-12-31') AND
(QA_US_MASTER.item_str in view
qa4_US_MASTER_VIEWS.item_str.ACTV_FLG = 'Y'))"). Spool 3 and
QA_US_MASTER.item_str are joined using a dynamic hash join, with a
join condition of ("QA_US_MASTER.item_str.STR_LOC_ID = LOC_ID").
The input table QA_US_MASTER.item_str will not be cached in memory.
The result goes into Spool 5 (all_amps) (compressed columns
allowed), which is built locally on the AMPs into 14 hash join
partitions. The size of Spool 5 is estimated with no confidence
to be 69,133,946 rows (2,419,688,110 bytes). The estimated time
for this step is 1 minute and 8 seconds.
4) We do an all-AMPs JOIN step from Spool 4 (Last Use) by way of an
all-rows scan, which is joined to Spool 5 (Last Use) by way of an
all-rows scan. Spool 4 and Spool 5 are joined using a hash join
of 14 partitions, with a join condition of ("(ITEM_ID = ITEM_ID)
AND (ACTV_FLG = ACTV_FLG)"). The result goes into Spool 6
(all_amps) (compressed columns allowed), which is redistributed by
the hash code of (QA_US_MASTER.item_str.STR_LOC_ID,
QA_US_MASTER.item_str.ITEM_STAT_CD, QA_US_MASTER.item_str.ITEM_ID,
QA_US_MASTER.ITEM.ITEM_ID, QA_US_MASTER.LOCATION.LOC_ID) to all
AMPs into 33 hash join partitions. The size of Spool 6 is
estimated with no confidence to be 36,434,893 rows (1,603,135,292
bytes). The estimated time for this step is 9.11 seconds.
5) We do an all-AMPs RETRIEVE step from QA_US_MASTER.MVNDR_ITEM_STR
in view qa4_US_MASTER_VIEWS.mvndr_item_str by way of an all-rows
scan with a condition of ("(QA_US_MASTER.MVNDR_ITEM_STR in view
qa4_US_MASTER_VIEWS.mvndr_item_str.CURR_DSVC_TYP_CD = 3) AND
((QA_US_MASTER.MVNDR_ITEM_STR in view
qa4_US_MASTER_VIEWS.mvndr_item_str.EFF_END_DT = DATE '9999-12-31')
AND ((QA_US_MASTER.MVNDR_ITEM_STR in view
qa4_US_MASTER_VIEWS.mvndr_item_str.ACTV_FLG = 'Y') AND
(QA_US_MASTER.MVNDR_ITEM_STR in view
qa4_US_MASTER_VIEWS.mvndr_item_str.OK_TO_ORD_FLG = 'Y')))") into
Spool 7 (all_amps) (compressed columns allowed) fanned out into 33
hash join partitions, which is redistributed by the hash code of (
QA_US_MASTER.MVNDR_ITEM_STR.ITEM_ID,
QA_US_MASTER.MVNDR_ITEM_STR.STR_LOC_ID,
QA_US_MASTER.MVNDR_ITEM_STR.ITEM_STAT_CD,
QA_US_MASTER.MVNDR_ITEM_STR.ITEM_ID,
QA_US_MASTER.MVNDR_ITEM_STR.STR_LOC_ID) to all AMPs. The input
table will not be cached in memory, but it is eligible for
synchronized scanning. The size of Spool 7 is estimated with no
confidence to be 173,967,551 rows (5,914,896,734 bytes). The
estimated time for this step is 2 minutes and 23 seconds.
6) We do an all-AMPs JOIN step from Spool 6 (Last Use) by way of an
all-rows scan, which is joined to Spool 7 (Last Use) by way of an
all-rows scan. Spool 6 and Spool 7 are joined using a hash join
of 33 partitions, with a join condition of ("(STR_LOC_ID =
STR_LOC_ID) AND ((ITEM_STAT_CD = ITEM_STAT_CD) AND ((ITEM_ID =
ITEM_ID) AND ((ACTV_FLG = OK_TO_ORD_FLG) AND ((ACTV_FLG = ACTV_FLG)
AND ((EFF_END_DT = EFF_END_DT) AND ((ACTV_FLG = ACTV_FLG) AND
((OK_TO_ORD_FLG = ACTV_FLG) AND ((ITEM_ID = ITEM_ID) AND
(STR_LOC_ID = LOC_ID )))))))))"). The result goes into Spool 2
(all_amps) (compressed columns allowed), which is built locally on
the AMPs. The size of Spool 2 is estimated with no confidence to
be 12,939,628 rows (336,430,328 bytes). The estimated time for
this step is 4.00 seconds.
7) We do an all-AMPs STAT FUNCTION step from Spool 2 by way of an
all-rows scan into Spool 10, which is redistributed by hash code
to all AMPs. The result rows are put into Spool 1 (group_amps),
which is built locally on the AMPs. This step is used to retrieve
the TOP 100 rows. Load distribution optimization is used.
If this step retrieves less than 100 rows, then execute step 8.
The size is estimated with no confidence to be 100 rows (3,200
bytes).
8) We do an all-AMPs STAT FUNCTION step from Spool 2 (Last Use) by
way of an all-rows scan into Spool 10 (Last Use), which is
redistributed by hash code to all AMPs. The result rows are put
into Spool 1 (group_amps), which is built locally on the AMPs.
This step is used to retrieve the TOP 100 rows. The size is
estimated with no confidence to be 100 rows (3,200 bytes).
9) Finally, we send out an END TRANSACTION step to all AMPs involved
in processing the request.
-> The contents of Spool 1 are sent back to the user as the result of
statement 1.
【问题讨论】：
-
在每个基表上粘贴解释计划和统计数据
-
整个查询的解释计划是什么？
-
为整个查询添加了解释计划
标签： query-optimization teradata