Question

在具有JOIN的Siddhi查询中，最好将条件条件放在ON子句或过滤条件中。

JOIN ON子句中的条件

FROM Stroke_TPA_Ingest_Ras_eMar_Output_Stream as Tpa
JOIN PreAdmitADTTable as Adt
  ON str:concat(Tpa.FacilityMnemonic,"-",Tpa.AccountNumber) == str:concat(Adt.FacilityMnemonic,"-",Adt.AccountNumber)
  AND (
        (
          (str:contains(Tpa.TpaPharmacyCodesValueSet, "https://fhir.app.medcity.net/ValueSet/StrokeTpaPharmacyCodes|"))
          AND 
          (convert(Tpa.AdministrationDosageFormId, 'int') > 40 AND convert(Tpa.AdministrationDosageFormId, 'int') <= 90)
        )
        OR 
        (
          (str:contains(Tpa.TpaPharmacyCodesValueSet, "https://fhir.app.medcity.net/ValueSet/StrokeTpaBolusPharmacyCodes|"))
          AND 
          (convert(Tpa.AdministrationDosageFormId, 'int') > 0 AND convert(Tpa.AdministrationDosageFormId, 'int') <= 9)
        )
      )
SELECT  Tpa.meta_LinkId  as meta_LinkId
INSERT INTO AlertStreamIntermediary;

过滤器中的条件

FROM Stroke_TPA_Ingest_Ras_eMar_Output_Stream[
      ((str:contains(Tpa.TpaPharmacyCodesValueSet, "https://fhir.app.medcity.net/ValueSet/StrokeTpaPharmacyCodes|"))
      AND 
      (convert(Tpa.AdministrationDosageFormId, 'int') > 40 AND convert(Tpa.AdministrationDosageFormId, 'int') <= 90))
    OR  
      ((str:contains(Tpa.TpaPharmacyCodesValueSet, "https://fhir.app.medcity.net/ValueSet/StrokeTpaBolusPharmacyCodes|"))
      AND
      (convert(Tpa.AdministrationDosageFormId, 'int') > 0 AND convert(Tpa.AdministrationDosageFormId, 'int') <= 9))
  ] as Tpa
JOIN PreAdmitADTTable as Adt
  ON str:concat(Tpa.FacilityMnemonic,"-",Tpa.AccountNumber) == str:concat(Adt.FacilityMnemonic,"-",Adt.AccountNumber)
SELECT  Tpa.meta_LinkId  as meta_LinkId
INSERT INTO AlertStreamIntermediary;

这两个查询似乎是等效的，但是想知道是否存在性能差异

Answer 1

第二个查询的性能会更好，因为当事件到达连接查询时，我们已经使用过滤器查询消除了不需要的事件。由于Siddhi用作管道，与过滤器不匹配的事件将被丢弃在该层。有关Siddhi体系结构的更多信息，您可以参考this。

有利于将条件设置为JOIN ON或Filter条件

1 个答案: