[
https://issues.apache.org/jira/browse/SOLR-13132?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17129705#comment-17129705
]
Michael Gibney commented on SOLR-13132:
---------------------------------------
Fair enough, perfectly fine to prefer consistency for allBuckets skg across
sweep/non-sweep ... I hesitated to mention it, but did anyway.
I will add javadocs to the "Shim" class (sorry about that); but to briefly
comment on the necessity to "shim" at all: the last merge from master
introduced {{DEV_NULL_SLOT_ACC}}, which, not being an instance of
{{SweepingCountSlotAcc}}, threw a {{ClassCastException}} on the hard cast in
{{collectDocs()}} (in {{ByArrayUIF}} and {{ByArrayDV}}). This pointed to a more
general problem, which was that if a subclass _were_ to set a {{countAcc}} that
doesn't support sweeping, it would similarly throw {{ClassCastException}} on
the hard cast in {{collectDocs()}}.
To support such a case, a shim is required because the code paths that do the
actual count accumulation (in {{ByArrayUIF}} and {{ByArrayDV}}) used to
directly increment {{processor.countAcc}}, and have now been switched to
register counts via the {{SweepDocIterator}} and {{SweepDISI}} abstractions,
respectively. So the goal of the "shim" is really just to construct one of
those {{Sweep*}} objects, to use the same codepaths to accumulate counts into
the non-sweep countAcc as a "sweep of one" -- it should be functionally
identical to current master (which directly increments {{processor.countAcc}}).
> Improve JSON "terms" facet performance when sorted by relatedness
> ------------------------------------------------------------------
>
> Key: SOLR-13132
> URL: https://issues.apache.org/jira/browse/SOLR-13132
> Project: Solr
> Issue Type: Improvement
> Components: Facet Module
> Affects Versions: 7.4, master (9.0)
> Reporter: Michael Gibney
> Priority: Major
> Attachments: SOLR-13132-with-cache-01.patch,
> SOLR-13132-with-cache.patch, SOLR-13132.patch, SOLR-13132_testSweep.patch
>
> Time Spent: 1.5h
> Remaining Estimate: 0h
>
> When sorting buckets by {{relatedness}}, JSON "terms" facet must calculate
> {{relatedness}} for every term.
> The current implementation uses a standard uninverted approach (either
> {{docValues}} or {{UnInvertedField}}) to get facet counts over the domain
> base docSet, and then uses that initial pass as a pre-filter for a
> second-pass, inverted approach of fetching docSets for each relevant term
> (i.e., {{count > minCount}}?) and calculating intersection size of those sets
> with the domain base docSet.
> Over high-cardinality fields, the overhead of per-term docSet creation and
> set intersection operations increases request latency to the point where
> relatedness sort may not be usable in practice (for my use case, even after
> applying the patch for SOLR-13108, for a field with ~220k unique terms per
> core, QTime for high-cardinality domain docSets were, e.g.: cardinality
> 1816684=9000ms, cardinality 5032902=18000ms).
> The attached patch brings the above example QTimes down to a manageable
> ~300ms and ~250ms respectively. The approach calculates uninverted facet
> counts over domain base, foreground, and background docSets in parallel in a
> single pass. This allows us to take advantage of the efficiencies built into
> the standard uninverted {{FacetFieldProcessorByArray[DV|UIF]}}), and avoids
> the per-term docSet creation and set intersection overhead.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
