BaseX query optimization on join
Question
After the issue in the following Stackoverflow is fixed, I have another problem when I try to make a join as below. The last query takes about 250ms while the first two take only 16ms. Is there a better way to perform join between two items?
Note: You can find the test data from this link .
let $PlGeTys :=
/root/PlGeTys/PlGeTy[
isOfPlCt/@href=/root/PlCts/PlCt[
environment='AIR'
]/@id
]
let $PlSpTys :=
/root/PlSpTys/PlSpTy[
isOfPlGeTy/@href=$PlGeTys/@id
]
for $PlGeTy in $PlGeTys,
$PlSpTy in $PlSpTys
where $PlSpTy/isOfPlGeTy/@href=$PlGeTy/@id
return <done>{$PlGeTy, $PlSpTy }</done>
Here is the query info:
Compiling:
- applying attribute index for $PlGeTys_0/@*:id
- rewriting where clause(s)
Query:
let $PlGeTys := /root/PlGeTys/PlGeTy[ isOfPlCt/@href=/root/PlCts/PlCt[ environment='AIR' ]/@id ] let $PlSpTys := /root/PlSpTys/PlSpTy[ isOfPlGeTy/@href=$PlGeTys/@id ] for $PlGeTy in $PlGeTys, $PlSpTy in $PlSpTys where $PlSpTy/isOfPlGeTy/@href=$PlGeTy/@id return "done"
Optimized Query:
let $PlGeTys_0 := db:open-pre("Output6",0)/*:root/*:PlGeTys/*:PlGeTy[(*:isOfPlCt/@*:href = root()/*:root/*:PlCts/*:PlCt[(*:environment = "AIR")]/@*:id)] let $PlSpTys_1 := db:attribute("Output6", $PlGeTys_0/@*:id)/self::*:href/parent::*:isOfPlGeTy/parent::*:PlSpTy for $PlGeTy_2 in $PlGeTys_0 for $PlSpTy_3 in ($PlSpTys_1)[(isOfPlGeTy/@href = $PlGeTy_2/@*:id)] return "done"
Result:
- Hit(s): 3642 Items
- Updated: 0 Items
- Printed: 18209 Bytes
- Read Locking: local [Output6]
- Write Locking: none
Timing:
- Parsing: 0.77 ms
- Compiling: 0.47 ms
- Evaluating: 215.71 ms
- Printing: 0.17 ms
- Total Time: 217.11 ms
Query plan:
<QueryPlan compiled="true">
<GFLWOR>
<Let>
<Var name="$PlGeTys" id="0"/>
<IterPath>
<DBNode name="Output6" pre="0"/>
<IterStep axis="child" test="*:root"/>
<IterStep axis="child" test="*:PlGeTys"/>
<IterStep axis="child" test="*:PlGeTy">
<CmpG op="=">
<CachedPath>
<IterStep axis="child" test="*:isOfPlCt"/>
<IterStep axis="attribute" test="*:href"/>
</CachedPath>
<IterPath>
<Root/>
<IterStep axis="child" test="*:root"/>
<IterStep axis="child" test="*:PlCts"/>
<IterStep axis="child" test="*:PlCt">
<CmpG op="=">
<CachedPath>
<IterStep axis="child" test="*:environment"/>
</CachedPath>
<Str value="AIR" type="xs:string"/>
</CmpG>
</IterStep>
<IterStep axis="attribute" test="*:id"/>
</IterPath>
</CmpG>
</IterStep>
</IterPath>
</Let>
<Let>
<Var name="$PlSpTys" id="1"/>
<CachedPath>
<ValueAccess data="Output6" type="ATTRIBUTE">
<CachedPath>
<VarRef>
<Var name="$PlGeTys" id="0"/>
</VarRef>
<IterStep axis="attribute" test="*:id"/>
</CachedPath>
</ValueAccess>
<IterStep axis="self" test="*:href"/>
<IterStep axis="parent" test="*:isOfPlGeTy"/>
<IterStep axis="parent" test="*:PlSpTy"/>
</CachedPath>
</Let>
<For>
<Var name="$PlGeTy" id="2"/>
<VarRef>
<Var name="$PlGeTys" id="0"/>
</VarRef>
</For>
<For>
<Var name="$PlSpTy" id="3"/>
<IterFilter>
<VarRef>
<Var name="$PlSpTys" id="1"/>
</VarRef>
<CmpG op="=">
<CachedPath>
<IterStep axis="child" test="isOfPlGeTy"/>
<IterStep axis="attribute" test="href"/>
</CachedPath>
<IterPath>
<VarRef>
<Var name="$PlGeTy" id="2"/>
</VarRef>
<IterStep axis="attribute" test="*:id"/>
</IterPath>
</CmpG>
</IterFilter>
</For>
<Str value="done" type="xs:string"/>
</GFLWOR>
</QueryPlan>
Update two:
The issue above is fixed. But when I want to use it in a local function as below, it takes about 700 ms. Am I doing something wrong?
declare function local:result($root as element(root)) as element()* {
let $PlCts := $root/PlCts/PlCt[environment = 'AIR']/@id
for $PlGeTy in $root/PlGeTys/PlGeTy[isOfPlCt/@href = $PlCts],
$PlSpTy in $root/PlSpTys/PlSpTy[isOfPlGeTy/@href = $PlGeTy/@id]
return <done>{ $PlGeTy, $PlSpTy }</done>
};
let $result := local:result(/root)
return $result
Query info;
Compiling:
- removing redundant element()* cast.
- inlining local:result#1
- inlining $root_5 as element(root)
- simplifying flwor expression
Query:
declare function local:result($root as element(root)) as element()* { let $PlCts := $root/PlCts/PlCt[environment = 'AIR']/@id for $PlGeTy in $root/PlGeTys/PlGeTy[isOfPlCt/@href = $PlCts], $PlSpTy in $root/PlSpTys/PlSpTy[isOfPlGeTy/@href = $PlGeTy/@id] return <done>{ $PlGeTy, $PlSpTy }</done> }; let $result := local:result(/root) return $result
Optimized Query:
let $PlCts_6 := ((: element(root), true :) db:open-pre("Output6",0)/*:root)/PlCts/PlCt[(environment = "AIR")]/@id let $result_4 := for $PlGeTy_7 in ((: element(root), true :) db:open-pre("Output6",0)/*:root)/PlGeTys/PlGeTy[(isOfPlCt/@href = $PlCts_6)] for $PlSpTy_8 in ((: element(root), true :) db:open-pre("Output6",0)/*:root)/PlSpTys/PlSpTy[(isOfPlGeTy/@href = $PlGeTy_7/@id)] return element done { (($PlGeTy_7, $PlSpTy_8)) } return $result_4
Result:
- Hit(s): 3642 Items
- Updated: 0 Items
- Printed: 553 KB
- Read Locking: local [Output6]
- Write Locking: none
Timing:
- Parsing: 1.41 ms
- Compiling: 2.9 ms
- Evaluating: 581.5 ms
- Printing: 8.34 ms
- Total Time: 594.15 ms
Query plan:
<QueryPlan compiled="true">
<GFLWOR>
<Let>
<Var name="$PlCts" id="6"/>
<IterPath>
<TypeCheck type="element(root)" function="true">
<IterPath>
<DBNode name="Output6" pre="0"/>
<IterStep axis="child" test="*:root"/>
</IterPath>
</TypeCheck>
<IterStep axis="child" test="PlCts"/>
<IterStep axis="child" test="PlCt">
<CmpG op="=">
<CachedPath>
<IterStep axis="child" test="environment"/>
</CachedPath>
<Str value="AIR" type="xs:string"/>
</CmpG>
</IterStep>
<IterStep axis="attribute" test="id"/>
</IterPath>
</Let>
<Let>
<Var name="$result" id="4"/>
<GFLWOR>
<For>
<Var name="$PlGeTy" id="7"/>
<IterPath>
<TypeCheck type="element(root)" function="true">
<IterPath>
<DBNode name="Output6" pre="0"/>
<IterStep axis="child" test="*:root"/>
</IterPath>
</TypeCheck>
<IterStep axis="child" test="PlGeTys"/>
<IterStep axis="child" test="PlGeTy">
<CmpG op="=">
<CachedPath>
<IterStep axis="child" test="isOfPlCt"/>
<IterStep axis="attribute" test="href"/>
</CachedPath>
<VarRef>
<Var name="$PlCts" id="6"/>
</VarRef>
</CmpG>
</IterStep>
</IterPath>
</For>
<For>
<Var name="$PlSpTy" id="8"/>
<IterPath>
<TypeCheck type="element(root)" function="true">
<IterPath>
<DBNode name="Output6" pre="0"/>
<IterStep axis="child" test="*:root"/>
</IterPath>
</TypeCheck>
<IterStep axis="child" test="PlSpTys"/>
<IterStep axis="child" test="PlSpTy">
<CmpG op="=">
<CachedPath>
<IterStep axis="child" test="isOfPlGeTy"/>
<IterStep axis="attribute" test="href"/>
</CachedPath>
<IterPath>
<VarRef>
<Var name="$PlGeTy" id="7"/>
</VarRef>
<IterStep axis="attribute" test="id"/>
</IterPath>
</CmpG>
</IterStep>
</IterPath>
</For>
<CElem>
<QNm value="done" type="xs:QName"/>
<List>
<VarRef>
<Var name="$PlGeTy" id="7"/>
</VarRef>
<VarRef>
<Var name="$PlSpTy" id="8"/>
</VarRef>
</List>
</CElem>
</GFLWOR>
</Let>
<VarRef>
<Var name="$result" id="4"/>
</VarRef>
</GFLWOR>
</QueryPlan>
Update three: Now, I have another issue when I extend the query with the additional "Cont" items, The query execution lasts about 600 ms. But without "Cont" items, it takes only 35 ms. Do you any suggestion on optimizing this query?
let $PlCts := /root/PlCts/PlCt[environment = 'AIR']/@id
for $PlGeTy in /root/PlGeTys/PlGeTy[isOfPlCt/@href = $PlCts],
$PlSpTy in /root/PlSpTys/PlSpTy[isOfPlGeTy/@href = $PlGeTy/@id],
$Cont in /root/Conts/Cont[@id=$PlSpTy/isOfCont/@href]
return <done>{ $PlGeTy, $PlSpTy, $Cont }</done>
This is the link of the latest XML data for testing.
The query info:
Compiling:
- applying text index for "AIR"
- applying attribute index for $PlCts_0
- applying attribute index for $PlGeTy_1/@id
- applying attribute index for $PlSpTy_2/isOfCont/@href
- inlining $PlCts_0
Query:
let $PlCts := /root/PlCts/PlCt[environment = 'AIR']/@id for $PlGeTy in /root/PlGeTys/PlGeTy[isOfPlCt/@href = $PlCts], $PlSpTy in /root/PlSpTys/PlSpTy[isOfPlGeTy/@href = $PlGeTy/@id], $Cont in /root/Conts/Cont[@id=$PlSpTy/isOfCont/@href] return <done>{ $PlGeTy, $PlSpTy, $Cont }</done>
Optimized Query:
for $PlGeTy_1 in db:attribute("Output7", db:text("Output7", "AIR")/parent::*:environment/parent::*:PlCt/@*:id)/self::*:href/parent::*:isOfPlCt/parent::*:PlGeTy for $PlSpTy_2 in db:attribute("Output7", $PlGeTy_1/@id)/self::*:href/parent::*:isOfPlGeTy/parent::*:PlSpTy for $Cont_3 in db:attribute("Output7", $PlSpTy_2/isOfCont/@href)/self::*:id/parent::*:Cont return element done { (($PlGeTy_1, $PlSpTy_2, $Cont_3)) }
Result:
- Hit(s): 3642 Items
- Updated: 0 Items
- Printed: 1159 KB
- Read Locking: local [Output7]
- Write Locking: none
Timing:
- Parsing: 0.39 ms
- Compiling: 0.68 ms
- Evaluating: 585.29 ms
- Printing: 14.36 ms
- Total Time: 600.72 ms
Query plan:
<QueryPlan compiled="true">
<GFLWOR>
<For>
<Var name="$PlGeTy" id="1"/>
<CachedPath>
<ValueAccess data="Output7" type="ATTRIBUTE">
<CachedPath>
<ValueAccess data="Output7" type="TEXT" name="*:environment">
<Str value="AIR" type="xs:string"/>
</ValueAccess>
<IterStep axis="parent" test="*:PlCt"/>
<IterStep axis="attribute" test="*:id"/>
</CachedPath>
</ValueAccess>
<IterStep axis="self" test="*:href"/>
<IterStep axis="parent" test="*:isOfPlCt"/>
<IterStep axis="parent" test="*:PlGeTy"/>
</CachedPath>
</For>
<For>
<Var name="$PlSpTy" id="2"/>
<CachedPath>
<ValueAccess data="Output7" type="ATTRIBUTE">
<IterPath>
<VarRef>
<Var name="$PlGeTy" id="1"/>
</VarRef>
<IterStep axis="attribute" test="id"/>
</IterPath>
</ValueAccess>
<IterStep axis="self" test="*:href"/>
<IterStep axis="parent" test="*:isOfPlGeTy"/>
<IterStep axis="parent" test="*:PlSpTy"/>
</CachedPath>
</For>
<For>
<Var name="$Cont" id="3"/>
<CachedPath>
<ValueAccess data="Output7" type="ATTRIBUTE">
<IterPath>
<VarRef>
<Var name="$PlSpTy" id="2"/>
</VarRef>
<IterStep axis="child" test="isOfCont"/>
<IterStep axis="attribute" test="href"/>
</IterPath>
</ValueAccess>
<IterStep axis="self" test="*:id"/>
<IterStep axis="parent" test="*:Cont"/>
</CachedPath>
</For>
<CElem>
<QNm value="done" type="xs:QName"/>
<List>
<VarRef>
<Var name="$PlGeTy" id="1"/>
</VarRef>
<VarRef>
<Var name="$PlSpTy" id="2"/>
</VarRef>
<VarRef>
<Var name="$Cont" id="3"/>
</VarRef>
</List>
</CElem>
</GFLWOR>
</QueryPlan>
1 answers
solution1
1 ACCPTED 2015-06-16 12:58:30
In those for loops, the results of the two "large" XPath expressions are already cached by the optimizer. Furthermore, you're actually comparing the href/id attributes twice.
Cleaning up the query removing this double effort reduces the execution time by about 90%.
let $PlCts := /root/PlCts/PlCt[environment = 'AIR']/@id
for $PlGeTy in /root/PlGeTys/PlGeTy[isOfPlCt/@href = $PlCts],
$PlSpTy in /root/PlSpTys/PlSpTy[isOfPlGeTy/@href = $PlGeTy/@id]
return <done>{ $PlGeTy, $PlSpTy }</done>
The technical post webpages of this site follow the CC BY-SA 4.0 protocol. If you need to reprint, please indicate the site URL or the original address.Any question please contact:yoyou2525@163.com.
Related Question
join query optimization
MySQL query optimization [join]
Query JOIN or not? (optimization)
MySQL query with join optimization
Mysql Join Query optimization
Query optimization - join algorithms
SQL JOIN query Optimization with subqueries
Hive query optimization with map join
MySql Query Optimization with LEFT JOIN
Mysql join with group by query optimization
Related Tags