I generate a “where used” index as an XML doc in a separate database.
I’m then doing lookups against this index like this where $node is the element I want to find the index entry for:
collection($rkDatabase)/doc-where-used-index/where-used-entry[string(@key) eq string(db:node-id($node))]
Where the markup is shown in [1] below.
Using prof:track() I’m measuring a consistent 0.036 seconds per lookup, which seems very slow. Db:node-id() is instantaneous so the cost must be in the XPath itself. I’ve tried different variations on the predicate but nothing makes a measurable difference.
There must be a more efficient way to do this kind of lookup or a better way to structure my index XML so that the lookup is optimized.
Or is this something that could be affected by my index settings at the database level?
Thanks,
Eliot
[1] Where-used index markup:
<doc-where-used-index> <where-used-entry key="9151416" tagname="map" class="map/map" id=""> <title> <ph keyref="workplace-safety-management"/> </title> <conrefs/> <topicrefs> <noderef node-id="672471" database="tokyo" tagname="mapref" baseuri="/tokyo/qa-sushma-sundareswaran.ditamap" href="workplace-service-delivery-core.ditamap"/> <noderef node-id="898684" database="tokyo" tagname="mapref" baseuri="/tokyo/qa-ankit-jaiswal.ditamap" href="workplace-service-delivery-core.ditamap"/> <noderef node-id="9524996" database="tokyo" tagname="mapref" baseuri="/tokyo/workplace-service-delivery-suite.ditamap" href="workplace-service-delivery-core.ditamap"/> </topicrefs> <doc> <noderef node-id="9151416" database="tokyo" tagname="map" baseuri="/tokyo/workplace-service-delivery-core.ditamap"/> </doc> <xrefs/> <referencing-maps> <noderef node-id="672452" database="tokyo" tagname="map" baseuri="/tokyo/qa-sushma-sundareswaran.ditamap"/> <noderef node-id="898667" database="tokyo" tagname="map" baseuri="/tokyo/qa-ankit-jaiswal.ditamap"/> <noderef node-id="9524980" database="tokyo" tagname="map" baseuri="/tokyo/workplace-service-delivery-suite.ditamap"/> </referencing-maps> </where-used-entry> … </doc-where-used-index>
_____________________________________________ Eliot Kimber Sr Staff Content Engineer O: 512 554 9368 M: 512 554 9368 servicenow.comhttps://www.servicenow.com LinkedInhttps://www.linkedin.com/company/servicenow | Twitterhttps://twitter.com/servicenow | YouTubehttps://www.youtube.com/user/servicenowinc | Facebookhttps://www.facebook.com/servicenow
On Fri, 2022-07-15 at 18:04 +0000, Eliot Kimber wrote:
collection($rkDatabase)/doc-where-used-index/where-used- entry[string(@key) eq string(db:node-id($node))]
Where the markup is shown in [1] below.
Using prof:track() I’m measuring a consistent 0.036 seconds per lookup,
could it be opening the collection that takes time? Or are there hundreds of thousands of where-used-entry elements, and, if so, do you have an attribute value index?
There are on the order of 50K index entry items.
I think the attribute index is probably the answer—for some reason I assumed that would be indexed by default but I see that it is not looking at the database details.
I’ll try enabling the index and see what happens.
Cheers,
E.
_____________________________________________ Eliot Kimber Sr Staff Content Engineer O: 512 554 9368 M: 512 554 9368 servicenow.comhttps://www.servicenow.com LinkedInhttps://www.linkedin.com/company/servicenow | Twitterhttps://twitter.com/servicenow | YouTubehttps://www.youtube.com/user/servicenowinc | Facebookhttps://www.facebook.com/servicenow
From: Liam R. E. Quin liam@fromoldbooks.org Date: Friday, July 15, 2022 at 1:22 PM To: Eliot Kimber eliot.kimber@servicenow.com, basex-talk@mailman.uni-konstanz.de basex-talk@mailman.uni-konstanz.de Subject: Re: [basex-talk] Possible to Speed Up This Lookup? [External Email]
On Fri, 2022-07-15 at 18:04 +0000, Eliot Kimber wrote:
collection($rkDatabase)/doc-where-used-index/where-used- entry[string(@key) eq string(db:node-id($node))]
Where the markup is shown in [1] below.
Using prof:track() I’m measuring a consistent 0.036 seconds per lookup,
could it be opening the collection that takes time? Or are there hundreds of thousands of where-used-entry elements, and, if so, do you have an attribute value index?
-- Liam Quin, https://urldefense.com/v3/__https://www.delightfulcomputing.com/__;!!N4vogdj...https://urldefense.com/v3/__https:/www.delightfulcomputing.com/__;!!N4vogdjhuJM!C1fj636Ib2y0kENz2AwiUJZj6IyBT5mrb5YE_kSoMeqI0fFs95OpyWpldUalcS47Lg7Dt-1d0gNC7xyQXLy9rZQ$ Available for XML/Document/Information Architecture/XSLT/ XSL/XQuery/Web/Text Processing/A11Y training, work & consulting. Barefoot Web-slave, antique illustrations: https://urldefense.com/v3/__http://www.fromoldbooks.org__;!!N4vogdjhuJM!C1fj...https://urldefense.com/v3/__http:/www.fromoldbooks.org__;!!N4vogdjhuJM!C1fj636Ib2y0kENz2AwiUJZj6IyBT5mrb5YE_kSoMeqI0fFs95OpyWpldUalcS47Lg7Dt-1d0gNC7xyQWGj0qnM$
Yes, turning on the attribute index definitely improves performa nce dramatically, which is no surprise. Not sure how I missed the need to expressly turn on these indexes. Doing that now.
Cheers,
E.
_____________________________________________ Eliot Kimber Sr Staff Content Engineer O: 512 554 9368 M: 512 554 9368 servicenow.comhttps://www.servicenow.com LinkedInhttps://www.linkedin.com/company/servicenow | Twitterhttps://twitter.com/servicenow | YouTubehttps://www.youtube.com/user/servicenowinc | Facebookhttps://www.facebook.com/servicenow
From: Eliot Kimber eliot.kimber@servicenow.com Date: Friday, July 15, 2022 at 3:01 PM To: Liam R. E. Quin liam@fromoldbooks.org, basex-talk@mailman.uni-konstanz.de basex-talk@mailman.uni-konstanz.de Subject: Re: [basex-talk] Possible to Speed Up This Lookup? There are on the order of 50K index entry items.
I think the attribute index is probably the answer—for some reason I assumed that would be indexed by default but I see that it is not looking at the database details.
I’ll try enabling the index and see what happens.
Cheers,
E.
_____________________________________________ Eliot Kimber Sr Staff Content Engineer O: 512 554 9368 M: 512 554 9368 servicenow.comhttps://www.servicenow.com LinkedInhttps://www.linkedin.com/company/servicenow | Twitterhttps://twitter.com/servicenow | YouTubehttps://www.youtube.com/user/servicenowinc | Facebookhttps://www.facebook.com/servicenow
From: Liam R. E. Quin liam@fromoldbooks.org Date: Friday, July 15, 2022 at 1:22 PM To: Eliot Kimber eliot.kimber@servicenow.com, basex-talk@mailman.uni-konstanz.de basex-talk@mailman.uni-konstanz.de Subject: Re: [basex-talk] Possible to Speed Up This Lookup? [External Email]
On Fri, 2022-07-15 at 18:04 +0000, Eliot Kimber wrote:
collection($rkDatabase)/doc-where-used-index/where-used- entry[string(@key) eq string(db:node-id($node))]
Where the markup is shown in [1] below.
Using prof:track() I’m measuring a consistent 0.036 seconds per lookup,
could it be opening the collection that takes time? Or are there hundreds of thousands of where-used-entry elements, and, if so, do you have an attribute value index?
-- Liam Quin, https://urldefense.com/v3/__https://www.delightfulcomputing.com/__;!!N4vogdj...https://urldefense.com/v3/__https:/www.delightfulcomputing.com/__;!!N4vogdjhuJM!C1fj636Ib2y0kENz2AwiUJZj6IyBT5mrb5YE_kSoMeqI0fFs95OpyWpldUalcS47Lg7Dt-1d0gNC7xyQXLy9rZQ$ Available for XML/Document/Information Architecture/XSLT/ XSL/XQuery/Web/Text Processing/A11Y training, work & consulting. Barefoot Web-slave, antique illustrations: https://urldefense.com/v3/__http://www.fromoldbooks.org__;!!N4vogdjhuJM!C1fj...https://urldefense.com/v3/__http:/www.fromoldbooks.org__;!!N4vogdjhuJM!C1fj636Ib2y0kENz2AwiUJZj6IyBT5mrb5YE_kSoMeqI0fFs95OpyWpldUalcS47Lg7Dt-1d0gNC7xyQWGj0qnM$
Actually, I think I spoke too soon.
This test:
return (db:option('attrindex'), for $node in $map return prof:track(db:open($rkDatabase)/doc-where-used-index/where-used-entry/@key[xs:integer(.) eq 9151416])?time ! util:formatTime(.)
Returns:
true 00.030 seconds
Where my current database context is the database named in $rkDatabase.
A small improvement (0.036 to 0.030) but not dramatic.
So I think I’m still missing some detail, probably a configuration issue.
Cheers,
E, _____________________________________________ Eliot Kimber Sr Staff Content Engineer O: 512 554 9368 M: 512 554 9368 servicenow.comhttps://www.servicenow.com LinkedInhttps://www.linkedin.com/company/servicenow | Twitterhttps://twitter.com/servicenow | YouTubehttps://www.youtube.com/user/servicenowinc | Facebookhttps://www.facebook.com/servicenow
From: Eliot Kimber eliot.kimber@servicenow.com Date: Friday, July 15, 2022 at 3:39 PM To: Liam R. E. Quin liam@fromoldbooks.org, basex-talk@mailman.uni-konstanz.de basex-talk@mailman.uni-konstanz.de Subject: Re: [basex-talk] Possible to Speed Up This Lookup? Yes, turning on the attribute index definitely improves performa nce dramatically, which is no surprise. Not sure how I missed the need to expressly turn on these indexes. Doing that now.
Cheers,
E.
_____________________________________________ Eliot Kimber Sr Staff Content Engineer O: 512 554 9368 M: 512 554 9368 servicenow.comhttps://www.servicenow.com LinkedInhttps://www.linkedin.com/company/servicenow | Twitterhttps://twitter.com/servicenow | YouTubehttps://www.youtube.com/user/servicenowinc | Facebookhttps://www.facebook.com/servicenow
From: Eliot Kimber eliot.kimber@servicenow.com Date: Friday, July 15, 2022 at 3:01 PM To: Liam R. E. Quin liam@fromoldbooks.org, basex-talk@mailman.uni-konstanz.de basex-talk@mailman.uni-konstanz.de Subject: Re: [basex-talk] Possible to Speed Up This Lookup? There are on the order of 50K index entry items.
I think the attribute index is probably the answer—for some reason I assumed that would be indexed by default but I see that it is not looking at the database details.
I’ll try enabling the index and see what happens.
Cheers,
E.
_____________________________________________ Eliot Kimber Sr Staff Content Engineer O: 512 554 9368 M: 512 554 9368 servicenow.comhttps://www.servicenow.com LinkedInhttps://www.linkedin.com/company/servicenow | Twitterhttps://twitter.com/servicenow | YouTubehttps://www.youtube.com/user/servicenowinc | Facebookhttps://www.facebook.com/servicenow
From: Liam R. E. Quin liam@fromoldbooks.org Date: Friday, July 15, 2022 at 1:22 PM To: Eliot Kimber eliot.kimber@servicenow.com, basex-talk@mailman.uni-konstanz.de basex-talk@mailman.uni-konstanz.de Subject: Re: [basex-talk] Possible to Speed Up This Lookup? [External Email]
On Fri, 2022-07-15 at 18:04 +0000, Eliot Kimber wrote:
collection($rkDatabase)/doc-where-used-index/where-used- entry[string(@key) eq string(db:node-id($node))]
Where the markup is shown in [1] below.
Using prof:track() I’m measuring a consistent 0.036 seconds per lookup,
could it be opening the collection that takes time? Or are there hundreds of thousands of where-used-entry elements, and, if so, do you have an attribute value index?
-- Liam Quin, https://urldefense.com/v3/__https://www.delightfulcomputing.com/__;!!N4vogdj...https://urldefense.com/v3/__https:/www.delightfulcomputing.com/__;!!N4vogdjhuJM!C1fj636Ib2y0kENz2AwiUJZj6IyBT5mrb5YE_kSoMeqI0fFs95OpyWpldUalcS47Lg7Dt-1d0gNC7xyQXLy9rZQ$ Available for XML/Document/Information Architecture/XSLT/ XSL/XQuery/Web/Text Processing/A11Y training, work & consulting. Barefoot Web-slave, antique illustrations: https://urldefense.com/v3/__http://www.fromoldbooks.org__;!!N4vogdjhuJM!C1fj...https://urldefense.com/v3/__http:/www.fromoldbooks.org__;!!N4vogdjhuJM!C1fj636Ib2y0kENz2AwiUJZj6IyBT5mrb5YE_kSoMeqI0fFs95OpyWpldUalcS47Lg7Dt-1d0gNC7xyQWGj0qnM$
Is there a performance difference when using db:attribute() instead[1]?
Tim
[1] https://docs.basex.org/wiki/Database_Module#db:attribute
Yes, that is 100x faster.
Clearly I need to slow down and RTFM a bit more closely.
Thanks,
E. _____________________________________________ Eliot Kimber Sr Staff Content Engineer O: 512 554 9368 M: 512 554 9368 servicenow.comhttps://www.servicenow.com LinkedInhttps://www.linkedin.com/company/servicenow | Twitterhttps://twitter.com/servicenow | YouTubehttps://www.youtube.com/user/servicenowinc | Facebookhttps://www.facebook.com/servicenow
From: Tim Thompson timathom@gmail.com Date: Friday, July 15, 2022 at 4:40 PM To: Eliot Kimber eliot.kimber@servicenow.com Cc: Liam R. E. Quin liam@fromoldbooks.org, basex-talk@mailman.uni-konstanz.de basex-talk@mailman.uni-konstanz.de Subject: Re: [basex-talk] Possible to Speed Up This Lookup? [External Email]
Is there a performance difference when using db:attribute() instead[1]?
Tim
[1] https://docs.basex.org/wiki/Database_Module#db:attributehttps://urldefense.com/v3/__https:/docs.basex.org/wiki/Database_Module*db:attribute__;Iw!!N4vogdjhuJM!Ca2zI2049S4xTcwSnRICwcUHLqboyNh8rMFfVglLT9-A39rtWpP0ztq55CticIRYCsegnqwW0D7StcFb-Z4lHQ$
-- Tim A. Thompson (he, him) Librarian for Applied Metadata Research Yale University Library
On Fri, Jul 15, 2022 at 5:21 PM Eliot Kimber <eliot.kimber@servicenow.commailto:eliot.kimber@servicenow.com> wrote: Actually, I think I spoke too soon.
This test:
return (db:option('attrindex'), for $node in $map return prof:track(db:open($rkDatabase)/doc-where-used-index/where-used-entry/@key[xs:integer(.) eq 9151416])?time ! util:formatTime(.)
Returns:
true 00.030 seconds
Where my current database context is the database named in $rkDatabase.
A small improvement (0.036 to 0.030) but not dramatic.
So I think I’m still missing some detail, probably a configuration issue.
Cheers,
E, _____________________________________________ Eliot Kimber Sr Staff Content Engineer O: 512 554 9368 M: 512 554 9368 servicenow.comhttps://www.servicenow.com LinkedInhttps://urldefense.com/v3/__https:/www.linkedin.com/company/servicenow__;!!N4vogdjhuJM!Ca2zI2049S4xTcwSnRICwcUHLqboyNh8rMFfVglLT9-A39rtWpP0ztq55CticIRYCsegnqwW0D7StcFPeRPeww$ | Twitterhttps://urldefense.com/v3/__https:/twitter.com/servicenow__;!!N4vogdjhuJM!Ca2zI2049S4xTcwSnRICwcUHLqboyNh8rMFfVglLT9-A39rtWpP0ztq55CticIRYCsegnqwW0D7StcGPMd9U7w$ | YouTubehttps://urldefense.com/v3/__https:/www.youtube.com/user/servicenowinc__;!!N4vogdjhuJM!Ca2zI2049S4xTcwSnRICwcUHLqboyNh8rMFfVglLT9-A39rtWpP0ztq55CticIRYCsegnqwW0D7StcHUl5mSEQ$ | Facebookhttps://urldefense.com/v3/__https:/www.facebook.com/servicenow__;!!N4vogdjhuJM!Ca2zI2049S4xTcwSnRICwcUHLqboyNh8rMFfVglLT9-A39rtWpP0ztq55CticIRYCsegnqwW0D7StcEdjcThVg$
From: Eliot Kimber <eliot.kimber@servicenow.commailto:eliot.kimber@servicenow.com> Date: Friday, July 15, 2022 at 3:39 PM To: Liam R. E. Quin <liam@fromoldbooks.orgmailto:liam@fromoldbooks.org>, basex-talk@mailman.uni-konstanz.demailto:basex-talk@mailman.uni-konstanz.de <basex-talk@mailman.uni-konstanz.demailto:basex-talk@mailman.uni-konstanz.de> Subject: Re: [basex-talk] Possible to Speed Up This Lookup? Yes, turning on the attribute index definitely improves performa nce dramatically, which is no surprise. Not sure how I missed the need to expressly turn on these indexes. Doing that now.
Cheers,
E.
_____________________________________________ Eliot Kimber Sr Staff Content Engineer O: 512 554 9368 M: 512 554 9368 servicenow.comhttps://www.servicenow.com LinkedInhttps://urldefense.com/v3/__https:/www.linkedin.com/company/servicenow__;!!N4vogdjhuJM!Ca2zI2049S4xTcwSnRICwcUHLqboyNh8rMFfVglLT9-A39rtWpP0ztq55CticIRYCsegnqwW0D7StcFPeRPeww$ | Twitterhttps://urldefense.com/v3/__https:/twitter.com/servicenow__;!!N4vogdjhuJM!Ca2zI2049S4xTcwSnRICwcUHLqboyNh8rMFfVglLT9-A39rtWpP0ztq55CticIRYCsegnqwW0D7StcGPMd9U7w$ | YouTubehttps://urldefense.com/v3/__https:/www.youtube.com/user/servicenowinc__;!!N4vogdjhuJM!Ca2zI2049S4xTcwSnRICwcUHLqboyNh8rMFfVglLT9-A39rtWpP0ztq55CticIRYCsegnqwW0D7StcHUl5mSEQ$ | Facebookhttps://urldefense.com/v3/__https:/www.facebook.com/servicenow__;!!N4vogdjhuJM!Ca2zI2049S4xTcwSnRICwcUHLqboyNh8rMFfVglLT9-A39rtWpP0ztq55CticIRYCsegnqwW0D7StcEdjcThVg$
From: Eliot Kimber <eliot.kimber@servicenow.commailto:eliot.kimber@servicenow.com> Date: Friday, July 15, 2022 at 3:01 PM To: Liam R. E. Quin <liam@fromoldbooks.orgmailto:liam@fromoldbooks.org>, basex-talk@mailman.uni-konstanz.demailto:basex-talk@mailman.uni-konstanz.de <basex-talk@mailman.uni-konstanz.demailto:basex-talk@mailman.uni-konstanz.de> Subject: Re: [basex-talk] Possible to Speed Up This Lookup? There are on the order of 50K index entry items.
I think the attribute index is probably the answer—for some reason I assumed that would be indexed by default but I see that it is not looking at the database details.
I’ll try enabling the index and see what happens.
Cheers,
E.
_____________________________________________ Eliot Kimber Sr Staff Content Engineer O: 512 554 9368 M: 512 554 9368 servicenow.comhttps://www.servicenow.com LinkedInhttps://urldefense.com/v3/__https:/www.linkedin.com/company/servicenow__;!!N4vogdjhuJM!Ca2zI2049S4xTcwSnRICwcUHLqboyNh8rMFfVglLT9-A39rtWpP0ztq55CticIRYCsegnqwW0D7StcFPeRPeww$ | Twitterhttps://urldefense.com/v3/__https:/twitter.com/servicenow__;!!N4vogdjhuJM!Ca2zI2049S4xTcwSnRICwcUHLqboyNh8rMFfVglLT9-A39rtWpP0ztq55CticIRYCsegnqwW0D7StcGPMd9U7w$ | YouTubehttps://urldefense.com/v3/__https:/www.youtube.com/user/servicenowinc__;!!N4vogdjhuJM!Ca2zI2049S4xTcwSnRICwcUHLqboyNh8rMFfVglLT9-A39rtWpP0ztq55CticIRYCsegnqwW0D7StcHUl5mSEQ$ | Facebookhttps://urldefense.com/v3/__https:/www.facebook.com/servicenow__;!!N4vogdjhuJM!Ca2zI2049S4xTcwSnRICwcUHLqboyNh8rMFfVglLT9-A39rtWpP0ztq55CticIRYCsegnqwW0D7StcEdjcThVg$
From: Liam R. E. Quin <liam@fromoldbooks.orgmailto:liam@fromoldbooks.org> Date: Friday, July 15, 2022 at 1:22 PM To: Eliot Kimber <eliot.kimber@servicenow.commailto:eliot.kimber@servicenow.com>, basex-talk@mailman.uni-konstanz.demailto:basex-talk@mailman.uni-konstanz.de <basex-talk@mailman.uni-konstanz.demailto:basex-talk@mailman.uni-konstanz.de> Subject: Re: [basex-talk] Possible to Speed Up This Lookup? [External Email]
On Fri, 2022-07-15 at 18:04 +0000, Eliot Kimber wrote:
collection($rkDatabase)/doc-where-used-index/where-used- entry[string(@key) eq string(db:node-id($node))]
Where the markup is shown in [1] below.
Using prof:track() I’m measuring a consistent 0.036 seconds per lookup,
could it be opening the collection that takes time? Or are there hundreds of thousands of where-used-entry elements, and, if so, do you have an attribute value index?
-- Liam Quin, https://urldefense.com/v3/__https://www.delightfulcomputing.com/__;!!N4vogdj...https://urldefense.com/v3/__https:/www.delightfulcomputing.com/__;!!N4vogdjhuJM!C1fj636Ib2y0kENz2AwiUJZj6IyBT5mrb5YE_kSoMeqI0fFs95OpyWpldUalcS47Lg7Dt-1d0gNC7xyQXLy9rZQ$ Available for XML/Document/Information Architecture/XSLT/ XSL/XQuery/Web/Text Processing/A11Y training, work & consulting. Barefoot Web-slave, antique illustrations: https://urldefense.com/v3/__http://www.fromoldbooks.org__;!!N4vogdjhuJM!C1fj...https://urldefense.com/v3/__http:/www.fromoldbooks.org__;!!N4vogdjhuJM!C1fj636Ib2y0kENz2AwiUJZj6IyBT5mrb5YE_kSoMeqI0fFs95OpyWpldUalcS47Lg7Dt-1d0gNC7xyQWGj0qnM$
On Fri, 2022-07-15 at 21:20 +0000, Eliot Kimber wrote:
return (db:option('attrindex'), for $node in $map return prof:track(db:open($rkDatabase)/doc-where-used- index/where-used-entry/@key[xs:integer(.) eq 9151416])?time ! util:formatTime(.)
can you take the db:open() call out and see how much faster the expression runs without it?
return (db:option('attrindex'), let $db := db:open($rkDatabase) return for $node in $map
return prof:track($db/doc-where-used-index/where-used-entry /@key[xs:integer(.) eq 9151416])?time ! util:formatTime(.)
(if i have that right)
I profiled the db:open() and collection() calls and they were basically instantaneous.
Moving to using db:attribute() and the index makes my process 100x times faster. Now I feel like a bit of an idiot for not having figured this out sooner.
Somebody needs to write the Optimizing BaseX book….
Cheers,
E.
_____________________________________________ Eliot Kimber Sr Staff Content Engineer O: 512 554 9368 M: 512 554 9368 servicenow.comhttps://www.servicenow.com LinkedInhttps://www.linkedin.com/company/servicenow | Twitterhttps://twitter.com/servicenow | YouTubehttps://www.youtube.com/user/servicenowinc | Facebookhttps://www.facebook.com/servicenow
From: BaseX-Talk basex-talk-bounces@mailman.uni-konstanz.de on behalf of Liam R. E. Quin liam@fromoldbooks.org Date: Friday, July 15, 2022 at 4:55 PM To: basex-talk@mailman.uni-konstanz.de basex-talk@mailman.uni-konstanz.de Subject: Re: [basex-talk] Possible to Speed Up This Lookup? [External Email]
On Fri, 2022-07-15 at 21:20 +0000, Eliot Kimber wrote:
return (db:option('attrindex'), for $node in $map return prof:track(db:open($rkDatabase)/doc-where-used- index/where-used-entry/@key[xs:integer(.) eq 9151416])?time ! util:formatTime(.)
can you take the db:open() call out and see how much faster the expression runs without it?
return (db:option('attrindex'), let $db := db:open($rkDatabase) return for $node in $map
return prof:track($db/doc-where-used-index/where-used-entry /@key[xs:integer(.) eq 9151416])?time ! util:formatTime(.)
(if i have that right)
-- Liam Quin, https://urldefense.com/v3/__https://www.delightfulcomputing.com/__;!!N4vogdj...https://urldefense.com/v3/__https:/www.delightfulcomputing.com/__;!!N4vogdjhuJM!Ge6_HxIH-uWYEol5ImGje0BVsQ_FKkbp1x9czcNl7d0t6BKvoQ1v0sENn6_kHvOQI0dfFmEig6LEfg3dDcIyNVA$ Available for XML/Document/Information Architecture/XSLT/ XSL/XQuery/Web/Text Processing/A11Y training, work & consulting. Barefoot Web-slave, antique illustrations: https://urldefense.com/v3/__http://www.fromoldbooks.org__;!!N4vogdjhuJM!Ge6_...https://urldefense.com/v3/__http:/www.fromoldbooks.org__;!!N4vogdjhuJM!Ge6_HxIH-uWYEol5ImGje0BVsQ_FKkbp1x9czcNl7d0t6BKvoQ1v0sENn6_kHvOQI0dfFmEig6LEfg3dk_syZfc$
Hi Eliot -
Others are much more proficient and knowledgeable, but I try the following:
1. ABI := Always Be Indexing :) 2. I think that Christian has expressed that, with indexes in play, many queries can experience significant optimizations by reversing the query pattern; eg starting with the attribute value and working back up from there, etc. 3. The optimized query view can be super helpful in this regard, too.
I'm sure there are other, better habits and strategies, I know (I just don't have them ingrained! 🙂). Other users will share, I'm sure.
Have a pleasant weekend! Best, Bridger
On Fri, Jul 15, 2022, 6:01 PM Eliot Kimber eliot.kimber@servicenow.com wrote:
I profiled the db:open() and collection() calls and they were basically instantaneous.
Moving to using db:attribute() and the index makes my process 100x times faster. Now I feel like a bit of an idiot for not having figured this out sooner.
Somebody needs to write the Optimizing BaseX book….
Cheers,
E.
*Eliot Kimber*
Sr Staff Content Engineer
O: 512 554 9368
M: 512 554 9368
servicenow.com https://www.servicenow.com
LinkedIn https://www.linkedin.com/company/servicenow | Twitter https://twitter.com/servicenow | YouTube https://www.youtube.com/user/servicenowinc | Facebook https://www.facebook.com/servicenow
*From: *BaseX-Talk basex-talk-bounces@mailman.uni-konstanz.de on behalf of Liam R. E. Quin liam@fromoldbooks.org *Date: *Friday, July 15, 2022 at 4:55 PM *To: *basex-talk@mailman.uni-konstanz.de < basex-talk@mailman.uni-konstanz.de> *Subject: *Re: [basex-talk] Possible to Speed Up This Lookup?
[External Email]
On Fri, 2022-07-15 at 21:20 +0000, Eliot Kimber wrote:
return (db:option('attrindex'), for $node in $map return prof:track(db:open($rkDatabase)/doc-where-used- index/where-used-entry/@key[xs:integer(.) eq 9151416])?time ! util:formatTime(.)
can you take the db:open() call out and see how much faster the expression runs without it?
return (db:option('attrindex'), let $db := db:open($rkDatabase) return for $node in $map
returnprof:track($db/doc-where-used-index/where-used-entry /@key[xs:integer(.) eq 9151416])?time ! util:formatTime(.)
(if i have that right)
-- Liam Quin, https://urldefense.com/v3/__https://www.delightfulcomputing.com/__;!!N4vogdj... https://urldefense.com/v3/__https:/www.delightfulcomputing.com/__;!!N4vogdjhuJM!Ge6_HxIH-uWYEol5ImGje0BVsQ_FKkbp1x9czcNl7d0t6BKvoQ1v0sENn6_kHvOQI0dfFmEig6LEfg3dDcIyNVA$ Available for XML/Document/Information Architecture/XSLT/ XSL/XQuery/Web/Text Processing/A11Y training, work & consulting. Barefoot Web-slave, antique illustrations: https://urldefense.com/v3/__http://www.fromoldbooks.org__;!!N4vogdjhuJM!Ge6_... https://urldefense.com/v3/__http:/www.fromoldbooks.org__;!!N4vogdjhuJM!Ge6_HxIH-uWYEol5ImGje0BVsQ_FKkbp1x9czcNl7d0t6BKvoQ1v0sENn6_kHvOQI0dfFmEig6LEfg3dk_syZfc$
On Fri, 2022-07-15 at 22:00 +0000, Eliot Kimber wrote:
I profiled the db:open() and collection() calls and they were basically instantaneous.
Moving to using db:attribute() and the index makes my process 100x times faster. Now I feel like a bit of an idiot for not having figured this out sooner.
Nah, optimization is tricksy. Glad it's working!
Somebody needs to write the Optimizing BaseX book….
Cheers,
E.
Eliot Kimber Sr Staff Content Engineer O: 512 554 9368 M: 512 554 9368 servicenow.comhttps://www.servicenow.com LinkedInhttps://www.linkedin.com/company/servicenow | Twitterhttps://twitter.com/servicenow | YouTubehttps://www.youtube.com/user/servicenowinc | Facebookhttps://www.facebook.com/servicenow
From: BaseX-Talk basex-talk-bounces@mailman.uni-konstanz.de on behalf of Liam R. E. Quin liam@fromoldbooks.org Date: Friday, July 15, 2022 at 4:55 PM To: basex-talk@mailman.uni-konstanz.de <basex-talk@mailman.uni-konstanz.de
Subject: Re: [basex-talk] Possible to Speed Up This Lookup? [External Email]
On Fri, 2022-07-15 at 21:20 +0000, Eliot Kimber wrote:
return (db:option('attrindex'), for $node in $map return prof:track(db:open($rkDatabase)/doc-where-used- index/where-used-entry/@key[xs:integer(.) eq 9151416])?time ! util:formatTime(.)
can you take the db:open() call out and see how much faster the expression runs without it?
return (db:option('attrindex'), let $db := db:open($rkDatabase) return for $node in $map
return prof:track($db/doc-where-used-index/where-used-entry /@key[xs:integer(.) eq 9151416])?time ! util:formatTime(.)
(if i have that right)
-- Liam Quin, https://urldefense.com/v3/__https://www.delightfulcomputing.com/__;!!N4vogdj... < https://urldefense.com/v3/__https:/www.delightfulcomputing.com/__;!!N4vogdjh...
Available for XML/Document/Information Architecture/XSLT/ XSL/XQuery/Web/Text Processing/A11Y training, work & consulting. Barefoot Web-slave, antique illustrations: https://urldefense.com/v3/__http://www.fromoldbooks.org__;!!N4vogdjhuJM!Ge6_... < https://urldefense.com/v3/__http:/www.fromoldbooks.org__;!!N4vogdjhuJM!Ge6_H...
Nah, optimization is tricksy.
True. Both for the optimizer and the developer. I can spot two indicators for …
db:open($rkDatabase)/doc-where-used-index/where-used-entry/@key[xs:integer(.) eq 9151416]
… not being rewritten for index access:
• If $rkDatabase is not static, it won’t be possible at compile time to check if the database contains the required index. • BaseX has no numeric index (not yet [1]), which is why the predicate can’t be rewritten.
Queries of the following pattern …
db:open('database')/doc-where-used-index/where-used-entry/@key[. eq '9151416']
… should definitely be optimized.
It’s sometimes possible to enforce the use of an index [2], but to simplify things, it’s just fine to explicitly address the index via db:text and db:attribute.
We have listed some more examples for index rewritings in our documentation [3]. I’ll need to extend the compilation, as various more rewriting patterns were implemented since then. For example, both '=' and 'eq' will now be rewritten if we can see at compile time that the rewriting of 'eq' won’t ignore errors that might be raised in its original form (more than single items to be compared; invalid input types; …).
[1] https://github.com/BaseXdb/basex/issues/2069 [2] https://docs.basex.org/wiki/Indexes#Enforce_Rewritings [3] https://docs.basex.org/wiki/XQuery_Optimizations#Index_Rewritings
basex-talk@mailman.uni-konstanz.de
-
Bridger Dyson-Smith -
Christian Grün -
Eliot Kimber -
Liam R. E. Quin -
Tim Thompson