MySQL性能非常差

PFB一个SQL查询我试图运行。最终输出返回大约60k行，但需要接近5小时才能运行。连接和类似的东西没有问题，我觉得我的查询需要大量优化。有人可以指点我正确的方向吗？MySQL性能非常差

SELECT 
    rapidview.name AS RapidView, 
    CASE 
     WHEN linktype.LINKNAME ="jira_subtask_link" 
     THEN sprintdest.name 
     ELSE sprint.name 
    END AS Sprint, 
    j.pkey AS CaseKey, 
    -- Sub task arent assigned sprint details, they are directly pulled from parent task, so that 
    -- logic is implemented here for pulling all sprint related info 
    CASE 
     WHEN linktype.LINKNAME ="jira_subtask_link" 
     THEN FROM_UNIXTIME(sprintdest.start_date/1000) 
     ELSE FROM_UNIXTIME(sprint.start_date/1000) 
    END AS SprintStartDate, 
    CASE 
     WHEN linktype.LINKNAME ="jira_subtask_link" 
     THEN FROM_UNIXTIME(sprintdest.END_DATE/1000) 
     ELSE FROM_UNIXTIME(sprint.END_DATE/1000) 
    END      AS SprintEndDate, 
    StoryPoints.numbervalue AS StoryPoint, 
    c.cname     AS Component, 
    it.pname    AS Type, 
    p.pname     AS Project, 
    iss.pname    AS Status, 
    dest.pkey    AS linkedissue, 
    dest.id     AS destid, 
    dest.created   AS linkedissuecreated, 
    (cglinkedissue.created) AS LinkedIssueClosedDate, 
    linktype.LINKNAME  AS LinkType, 
    cfoowner.customvalue AS Owner, 
    j.created    AS Created, 
    cg.created    AS ClosedDate, 
    CASE 
     WHEN linktype.LINKNAME ="jira_subtask_link" 
     THEN (
       CASE 
        WHEN sprintdest.started=true 
        AND sprintdest.closed=false 
        THEN "Current Sprint" 
        WHEN sprintdest.started=true 
        AND sprintdest.closed=true 
        THEN "Completed Sprint" 
        WHEN sprintdest.started=false 
        AND sprintdest.closed=false 
        THEN "Future Sprint" 
       END) 
     ELSE (
       CASE 
        WHEN sprint.started=true 
        AND sprint.closed=false 
        THEN "Current Sprint" 
        WHEN sprint.started=true 
        AND sprint.closed=true 
        THEN "Completed Sprint" 
        WHEN sprint.started=false 
        AND sprint.closed=false 
        THEN "Future Sprint" 
       END) 
    END       AS SprintStatus, 
    j.TIMEORIGINALESTIMATE/3600 AS EstimatedTime, 
    j.TIMEESTIMATE/3600   AS RemainingTime, 
    j.TIMESPENT/3600   AS LoggedHours , 
    cg.id      AS CGID, 
    ci.groupid     AS cigroupid, 
    ci.field     AS CIFIELD, 
    ci.newstring    AS NEWSTRING 
    -- DevLead.stringvalue as DevLead, 
    -- PMLead.stringvalue as PMLead, 
    -- QaLead.stringvalue as QALead, 
    -- DevLeadName.display_name as DevleadDisplayName, 
    -- PMLeadName.display_name as PMLeadDisplayName, 
    -- QALeadName.display_name as QALeadDisplayName 
FROM 
    jiraissue j 
LEFT JOIN 
    customfieldvalue cfv 
ON 
    cfv.issue=j.id 
AND cfv.customfield=11002 
LEFT JOIN 
    AO_60DB71_SPRINT sprint 
ON 
    sprint.id=cfv.stringvalue 
LEFT JOIN 
    AO_60DB71_RAPIDVIEW rapidview 
ON 
    sprint.RAPID_VIEW_ID=rapidview.id 
LEFT JOIN 
    nodeassociation na 
ON 
    j.id=na.source_node_id 
AND na.association_type = ('IssueComponent') 
LEFT JOIN 
    component c 
ON 
    na.sink_node_id=c.id 
LEFT JOIN 
    customfieldvalue StoryPoints 
ON 
    j.id=StoryPoints.issue 
AND StoryPoints.customfield=10572 
    /* 
    LEFT JOIN 
    customfieldvalue PMLead 
    ON 
    j.id=PMLead.issue 
    AND PMLead.customfield=10382 
    LEFT JOIN 
    customfieldvalue DevLead 
    ON 
    j.id=DevLead.issue 
    AND StoryPoints.customfield=10380 
    LEFT JOIN 
    customfieldvalue QaLead 
    ON 
    j.id=QaLead.issue 
    AND QaLead.customfield=10381 
    left join cwd_user DevLeadName 
    on DevLead.stringvalue=DevLeadName.user_name 
    left join cwd_user PMLeadName 
    on PMLead.stringvalue=PMLeadName.user_name 
    left join cwd_user QALeadName 
    on QaLead.stringvalue=QALeadName.user_name 
    */ 
LEFT JOIN 
    issuetype it -- To pull in issuetype 
ON 
    j.issuetype=it.id 
LEFT JOIN 
    project p -- To pull in project 
ON 
    j.project=p.id 
LEFT JOIN 
    issuestatus iss -- To pull in Case Status 
ON 
    j.issuestatus=iss.id 
LEFT JOIN 
    issuelink il -- To identify linked cases 
ON 
    j.id=il.destination 
LEFT JOIN 
    issuelinktype linktype 
ON 
    il.linktype=linktype.id 
LEFT JOIN 
    jiraissue dest -- To idenfity component for the the linked case 
ON 
    dest.id=il.source 
LEFT JOIN 
    customfieldvalue owner -- To pull in customfields 
ON 
    j.id=owner.issue 
AND owner.customfield=10310 
LEFT JOIN 
    customfieldoption cfoowner -- To pull in customfields 
ON 
    cfoowner.id=owner.stringvalue 
LEFT JOIN 
    changegroup cg -- To pull in case history to identify status changes 
ON 
    j.id=cg.issueid 
LEFT JOIN 
    changeitem ci 
ON 
    cg.id=ci.groupid 
AND ci.field='status' 
AND ci.newstring LIKE '%Closed%' 
LEFT JOIN 
    changegroup cglinkedissue -- To pull in case history to identify status changes 
ON 
    dest.id=cglinkedissue.issueid 
LEFT JOIN 
    changeitem cilinkedissue 
ON 
    cilinkedissue.groupid=cglinkedissue.id 
AND cilinkedissue.field='status' 
AND cilinkedissue.newstring LIKE '%Closed%' 
LEFT JOIN 
    customfieldvalue cfvdest 
ON 
    cfvdest.issue=dest.id 
AND cfvdest.customfield=11002 
LEFT JOIN 
    AO_60DB71_SPRINT sprintdest 
ON 
    sprintdest.id=cfvdest.stringvalue 
    -- year(FROM_UNIXTIME(sprint.END_DATE/1000) /1000)>=2015 
    -- or year(FROM_UNIXTIME(sprintdest.END_DATE/1000) /1000)>=2015 
    -- where 
    -- j.pkey='CLQ-41441' 
    group by 
    j.id, 
    c.id,il.id,sprint.id

执行计划

id select_type table type possible_keys key key_len ref rows Extra 
1 SIMPLE j ALL (null) (null) (null) (null) 891945 (null) 
1 SIMPLE cfv ref cfvalue_issue cfvalue_issue 18 jira_rnd_p.j.ID,const 1 (null) 
1 SIMPLE sprint eq_ref PRIMARY PRIMARY 8 jira_rnd_p.cfv.STRINGVALUE 1 Using where 
1 SIMPLE rapidview eq_ref PRIMARY PRIMARY 8 jira_rnd_p.sprint.RAPID_VIEW_ID 1 (null) 
1 SIMPLE na ref PRIMARY,node_source PRIMARY 8 jira_rnd_p.j.ID 1 Using where; Using index 
1 SIMPLE c eq_ref PRIMARY PRIMARY 8 jira_rnd_p.na.SINK_NODE_ID 1 (null) 
1 SIMPLE StoryPoints ref cfvalue_issue cfvalue_issue 18 jira_rnd_p.j.ID,const 1 (null) 
1 SIMPLE it eq_ref PRIMARY PRIMARY 182 jira_rnd_p.j.issuetype 1 Using where 
1 SIMPLE p eq_ref PRIMARY PRIMARY 8 jira_rnd_p.j.PROJECT 1 (null) 
1 SIMPLE iss eq_ref PRIMARY PRIMARY 182 jira_rnd_p.j.issuestatus 1 Using where 
1 SIMPLE il ref issuelink_dest issuelink_dest 9 jira_rnd_p.j.ID 1 (null) 
1 SIMPLE linktype eq_ref PRIMARY PRIMARY 8 jira_rnd_p.il.LINKTYPE 1 (null) 
1 SIMPLE dest eq_ref PRIMARY PRIMARY 8 jira_rnd_p.il.SOURCE 1 (null) 
1 SIMPLE owner ref cfvalue_issue cfvalue_issue 18 jira_rnd_p.j.ID,const 1 (null) 
1 SIMPLE cfoowner eq_ref PRIMARY PRIMARY 8 jira_rnd_p.owner.STRINGVALUE 1 Using where 
1 SIMPLE cg ref chggroup_issue chggroup_issue 9 jira_rnd_p.j.ID 4 (null) 
1 SIMPLE ci ref chgitem_chggrp,chgitem_field chgitem_chggrp 9 jira_rnd_p.cg.ID 1 Using where 
1 SIMPLE cglinkedissue ref chggroup_issue chggroup_issue 9 jira_rnd_p.dest.ID 4 (null) 
1 SIMPLE cilinkedissue ref chgitem_chggrp,chgitem_field chgitem_chggrp 9 jira_rnd_p.cglinkedissue.ID 1 Using where 
1 SIMPLE cfvdest ref cfvalue_issue cfvalue_issue 18 jira_rnd_p.dest.ID,const 1 (null) 
1 SIMPLE sprintdest eq_ref PRIMARY PRIMARY 8 jira_rnd_p.cfvdest.STRINGVALUE 1 Using where

来源

2015-10-30 Adithya Kumar

那么，这是很多联接。查看执行计划，查看索引可能的帮助。 – Thilo

请原谅我格式化，因为我是新的发布问题的stackoverflow！ –

Hey Thilo，已经粘贴了执行计划。你可以帮助，因为我是新的SQL调优 –

尝试EXPLAIN。

请注意possible_keys,key,rows。

也许你可以发布EXPLAIN结果，我们可以看到该怎么做。

来源

2015-10-30 02:53:52

谢谢比利，我已经添加了解释结果。 –

你需要LEFT？也就是说，所有这些其他表是可选的吗？

如果在某些情况下您可以摆脱LEFT，则可以避免扫描所有891K行的j。

您是否只对“封闭”产品感兴趣？如果是这样，查询不会限制它们，因为LEFT。

我会开始删除LEFT只要可行。然后将末尾不是JOIN的子句的AND子句移动到WHERE。这可能允许查询更早地过滤东西，而不是在进入GROUP BY之前拖动891K（或更多）行（包括大量NULLs）。

来源

2016-01-25 21:25:35

MySQL性能非常差

回答

相关问题