+ All Categories
Home > Documents > Russ Houberg - SharePoint 2013 - Search Architecture

Russ Houberg - SharePoint 2013 - Search Architecture

Date post: 14-Oct-2015
Category:
Upload: kartik-anand
View: 33 times
Download: 0 times
Share this document with a friend
Description:
Great Article
Popular Tags:

of 30

Transcript
  • Presentedby:

  • Introduction

    RussHoubergSharePointMCSMKnowledgeLakeProfessionalServices

    [email protected]/rhoubergtwitter/rhouberg

  • Up for discussion

    Abriefhistory Concepts:SP2010 SP2013 Concepts:NewFunctionality FASTItsinthere ComponentTopologyOverhaul SchemaManagement PowerShellItsUnavoidable! ScalableArchitectureExamples ECMImpact Questions?

  • SharePoint2007 FASTESP

    CustomConnector

    SharePoint2010 FASTSearchforSharePoint2010

    SearchSSAisFASTAware TheBypass:EnterpriseSearch FS4SP EnterpriseSearch

    SharePoint2013 EnterpriseSearchw/FASTDNA

    Hybridvigorbringsthebestofbothworlds

    EnterpriseSearch(andECM)arestillgoingtobegenerallyonpremise(notcloud)

    Hybridvigor

    A brief history

  • Concepts: SP2010 SP2013Translating(orreplacing)thesyntax

    SharePoint2010 SharePoint2013

    ManagedProperty (Multiple) SearchSchemas

    BestBets PromotedResults(QueryRule)

    Scope andFederatedLocation ResultSource

    ContentByQuery Content BySearch

    Incremental Crawl ContinuousCrawl

    Mirrored IndexPartition IndexReplica

    MCM MCSM

  • Concepts: New FunctionalityNowwithbetterstuff

  • Concepts: New Functionality

    CrawlingProcessingChanges

    WenowhaveFull,IncrementalandContinuousCrawls

    NomoreHostDistributionRules

    NomoreFASTBypass

    Nowwithbetterstuff

  • Concepts: New Functionality

    ContinuousCrawlBenefits

    Nomorewaitingforindexmerge

    Doesnotwaitforothercrawlstocomplete

    Canhavemultiplecontinuouscrawlsrunningsimultaneously

    Continuouscrawlsignoreserrors

    Continus CrawlFacts

    Runsevery15minutesbydefault

    DefaultintervalcanbechangedwithPowerShell

    ShouldbeusedinsteadofincrementalcrawlsforSharePointcontentsources

    Nowwithbetterstuff

  • Concepts: New Functionality

    OtherBackendChanges

    Schemamanagementatsitecollectionlevel

    Crawldatabaserebalancing

    ResultSources&RemoteSPIndex

    Nowwithbetterstuff

  • Concepts: New Functionality

    QueryRules QueryRuletoimplement

    PromotedResults(bestbets) QueryRuletoaugmentsearch

    resultswithentitymetadata(fromSPorBCS)

    QueryRuletomanipulateresultsranking

    Nowwithbetterstuff

  • Concepts: New Functionality

    DEMOContentBySearch WebPart

    Nowwithbetterstuff

  • FAST Its in thereFASTbitsarenowblendedin

  • FAST Its in there

    ComponentsoftheFASTcrawlprocessing,indexstructure,relevanceandrankingmodels,entityextraction,andpipelineextensibilityoptions(throughwebservices)areallinfluencesofFASTtechnology.

    FASTbitsarenowblendedin

  • Component Topology OverhaulBreakingoutoftheshell

  • Component Topology Overhaul

    WhatitDoes

    Crawlscontentsourcestopopulateindex

    Deliverscrawlitems(binary)andmetadatatocontentprocessor

    Invokesconnectorsorprotocolhandlerstointeractwithcontentsourcestoretrievedata

    Usesoneormorecrawldatabasestostoreinfoaboutcrawlitemsandcrawlhistory

    ImportantFacts

    Wecanhavemultiplecrawlcomponents

    MSRecommends:2CrawlComponentsperSearchServiceApplication

    MSRecommends:8(4vm)CPU/8GBRAMperCrawlComponent

    CrawlComponent

  • Component Topology Overhaul

    WhatitDoes

    Processescrawlitemsandfeedstoindexcomponent

    Transformscrawlitemsintoartifactsthatcanbeincludedinsearchindex(Performsdocumentparsingandpropertymapping)

    Writesinformationaboutlinksandurls inlinkdatabase(whichareanalyzedbyanalyticstocalculaterelevanceandcurrency Resultswrittenbacktosearchindexbycontentprocessingcomponent

    Generatesphoneticnamevariationstoimprovepeoplesearch

    ImportantFacts

    Wemustonlyhaveone(1)crawlprocessingcomponentperserver morewillhurt,nothelpcrawlperformance

    Maxof2persearchserviceapplication FeedingSessionsarescaledbasedonCPUcoresusingadefaultcoefficientof3

    8(cores)*3=24feedingsessions4(cores)*3=12feedingsessions

    MSRecommends:8(4vm)CPU/8GBRAMperContentProcessingComponent

    FeedingsessionsrequireRAM MoreRAMisnecessarywhenmorecoresarepresent monitoringrequired

    ContentProcessingComponent

  • Component Topology Overhaul

    WhatitDoes Runsanalyticsjobsthatanalyzecrawlitemsanduserinteractionwithsearchresultstoperformbothsearchanalyticsandusageanalytics

    AnalyzesLink&Anchortextanalysis,Cleardistance,SearchClicks,DeepLinks,SocialTags,SocialDistance,SearchReports,Recommendations,UsageCounts,ActivityRanking

    Improvessearchrelevanceandcreatesearchresults

    Outputincludedinsearchindexbycontentprocessor

    ImportantFacts Maximumof6persearchserviceapplication AddmoreAnalyticsProcessingComponentstoimproveanalyticsperformance

    MSRecommends:8(4vm)CPU/8GBRAM/300GBdiskspaceperAnalyticsProcessingComponent.

    InteractswithAnalyticsReportingtostorestatisticalinformation

    InteractswithLinkdatabasetostoreinformationaboutsearchesandcrawleddocuments

    AnalyticsProcessingComponent

  • Component Topology Overhaul

    WhatitDoes

    Receivesprocesseditemsfromcontentprocessingcomponentandwritestheitemstotheindexfile

    Receivesqueriesfromthequeryprocessingcomponentandreturnsresultsets

    RedistributescontentamongindexpartitionswhenindexarchitectureischangedbySearchAdministrationComponent

    ImportantFacts

    Maximumof60indexpartitions(20indexpartitionsX3indexreplicas)persearchserviceapplication

    MustprovisiononeIndexComponentforeachindexreplica.

    MSRecommends:8(4vm)CPU/16GBRAM/500GBdiskspaceperIndexComponent.

    IndexComponent

  • Component Topology Overhaul

    Indexpartitionislogicalportionofentiresearchindex(sameasbefore)

    Indexpartitionisservedbyoneormoreindexcomponents Indexcomponentscanbeprimary"replica"orsecondary"replica" PrimaryReplicaiscontactedbycontentprocessingcomponenttowritenewdataintheindex

    SecondaryReplicaisreadonlycopythatgetupdatedwiththedata. Addingreplicasimprovesqueryperformanceunderload Addpartitionstohandleincreasedcontentcorpus Can'tremovepartitionafterithasbeenadded.

    IndexArchitecture

    IndexPartitionsandReplicas

  • Component Topology Overhaul

    WhatitDoes

    Analyzesandprocessesqueriesandresults

    Afterreceivingaquery,itanalyzesandprocessesthequerytooptimizeprecision,recallandrelevance

    Submitsprocessedqueriestotheindexcomponent

    Processestheresultsetreturnedbytheindexcomponentbeforereturningtothequeryingentity.

    ImportantFacts

    Maximumof1perserver MSRecommends:8(4vm)CPU/8GBRAMperQueryProcessingComponent.

    QueryProcessingComponent

  • Schema Management Schemachanges Granularconfiguration Canbeextendedatthesite

    collectionlevelormanagedgloballybyfarmadmin

    Fullcrawlneededtopickupchanges

    ItsallaboutSchemanow

  • PowerShell Its unavoidableYournewbestfriend

    MINIDEMOWhatyoucantdowiththeUI

  • PowerShell Its unavoidableGetSPEnterpriseSearchStatusNewSPEnterpriseSearchAdminComponentGetSPEnterpriseSearchCrawlContentSourceGetSPEnterpriseSearchCrawlCustomConnectorGetSPEnterpriseSearchCrawlDatabaseGetSPEnterpriseSearchCrawlExtensionGetSPEnterpriseSearchCrawlMappingGetSPEnterpriseSearchCrawlRuleNewSPEnterpriseSearchCrawlComponentNewSPEnterpriseSearchCrawlContentSourceNewSPEnterpriseSearchCrawlCustomConnectorNewSPEnterpriseSearchCrawlDatabaseNewSPEnterpriseSearchCrawlExtensionNewSPEnterpriseSearchCrawlMappingNewSPEnterpriseSearchCrawlRuleRemoveSPEnterpriseSearchCrawlContentSourceRemoveSPEnterpriseSearchCrawlCustomConnectorRemoveSPEnterpriseSearchCrawlDatabaseRemoveSPEnterpriseSearchCrawlExtensionRemoveSPEnterpriseSearchCrawlMappingRemoveSPEnterpriseSearchCrawlRuleSetSPEnterpriseSearchCrawlContentSourceSetSPEnterpriseSearchCrawlDatabaseSetSPEnterpriseSearchCrawlRuleSetSPEnterpriseSearchCrawlLogReadPermissionRemoveSPEnterpriseSearchCrawlLogReadPermissionRemoveSPEnterpriseSearchCrawlLogReadPermissionImportSPEnterpriseSearchTopologyExportSPEnterpriseSearchTopologySetSPEnterpriseSearchTopologyRemoveSPEnterpriseSearchTopologyRemoveSPEnterpriseSearchComponentNewSPEnterpriseSearchTopologyNewSPEnterpriseSearchQueryProcessingComponentNewSPEnterpriseSearchIndexComponent

    NewSPEnterpriseSearchContentProcessingComponentGetSPEnterpriseSearchTopologyGetSPEnterpriseSearchServiceGetSPEnterpriseSearchServiceApplicationGetSPEnterpriseSearchServiceApplicationProxyGetSPEnterpriseSearchServiceInstanceNewSPEnterpriseSearchServiceApplicationNewSPEnterpriseSearchServiceApplicationProxyRemoveSPEnterpriseSearchServiceApplicationRemoveSPEnterpriseSearchServiceApplicationProxyRestoreSPEnterpriseSearchServiceApplicationResumeSPEnterpriseSearchServiceApplicationSetSPEnterpriseSearchServiceSetSPEnterpriseSearchServiceApplicationSetSPEnterpriseSearchServiceApplicationProxyStartSPEnterpriseSearchServiceInstanceStopSPEnterpriseSearchServiceInstanceSuspendSPEnterpriseSearchServiceApplicationUpgradeSPEnterpriseSearchServiceApplicationBackupSPEnterpriseSearchServiceApplicationIndexUpgradeSPEnterpriseSearchServiceApplicationSiteSettingsRestoreSPEnterpriseSearchServiceApplicationIndexRemoveSPEnterpriseSearchServiceApplicationSiteSettingsGetSPEnterpriseSearchOwnerSuspendSPEnterpriseSearchServiceApplicationSetSPEnterpriseSearchServiceInstanceGetSPEnterpriseSearchMetadataCategoryGetSPEnterpriseSearchMetadataCrawledPropertyGetSPEnterpriseSearchMetadataManagedPropertyGetSPEnterpriseSearchMetadataMappingNewSPEnterpriseSearchMetadataCategoryNewSPEnterpriseSearchMetadataCrawledPropertyNewSPEnterpriseSearchMetadataManagedPropertyNewSPEnterpriseSearchMetadataMappingRemoveSPEnterpriseSearchMetadataCategoryRemoveSPEnterpriseSearchMetadataManagedPropertyRemoveSPEnterpriseSearchMetadataMappingSetSPEnterpriseSearchMetadataCategory

    SetSPEnterpriseSearchMetadataCrawledPropertySetSPEnterpriseSearchMetadataManagedPropertySetSPEnterpriseSearchMetadataMappingGetSPEnterpriseSearchPropertyRuleCollectionGetSPEnterpriseSearchPropertyRuleGetSPEnterpriseSearchQueryAndSiteSettingsServiceGetSPEnterpriseSearchQueryAndSiteSettingsServiceInstanceGetSPEnterpriseSearchQueryAndSiteSettingsServiceProxyGetSPEnterpriseSearchQueryAuthorityGetSPEnterpriseSearchQueryDemotedGetSPEnterpriseSearchQueryKeywordGetSPEnterpriseSearchQueryScopeGetSPEnterpriseSearchQueryScopeRuleGetSPEnterpriseSearchQuerySuggestionCandidatesGetSPEnterpriseSearchRankingModelGetSPEnterpriseSearchSecurityTrimmerNewSPEnterpriseSearchQueryAuthorityNewSPEnterpriseSearchQueryDemotedNewSPEnterpriseSearchQueryKeywordNewSPEnterpriseSearchQueryScopeNewSPEnterpriseSearchQueryScopeRuleNewSPEnterpriseSearchRankingModelNewSPEnterpriseSearchSecurityTrimmerRemoveSPEnterpriseSearchQueryAuthorityRemoveSPEnterpriseSearchQueryDemotedRemoveSPEnterpriseSearchQueryKeywordRemoveSPEnterpriseSearchQueryScopeRemoveSPEnterpriseSearchQueryScopeRuleRemoveSPEnterpriseSearchRankingModelRemoveSPEnterpriseSearchSecurityTrimmerSetSPEnterpriseSearchQueryAuthoritySetSPEnterpriseSearchQueryKeywordSetSPEnterpriseSearchQueryScopeSetSPEnterpriseSearchQueryScopeRuleSetSPEnterpriseSearchRankingModelStartSPEnterpriseSearchQueryAndSiteSettingsServiceInstanceStopSPEnterpriseSearchQueryAndSiteSettingsServiceInstanceImportSPEnterpriseSearchPopularQueries

    SetSPEnterpriseSearchResultItemTypeSetSPEnterpriseSearchQuerySpellingCorrectionRemoveSPEnterpriseSearchResultItemTypeNewSPEnterpriseSearchResultItemTypeImportSPEnterpriseSearchThesaurusGetSPEnterpriseSearchResultItemTypeGetSPEnterpriseSearchQuerySpellingCorrectionGetSPEnterpriseSearchLanguageResourcePhraseGetSPEnterpriseSearchSiteHitRuleNewSPEnterpriseSearchLanguageResourcePhraseNewSPEnterpriseSearchSiteHitRuleRemoveSPEnterpriseSearchLanguageResourcePhraseRemoveSPEnterpriseSearchSiteHitRuleGetSPEnterpriseSearchVssDataPathGetSPEnterpriseSearchContentEnrichmentConfigurationSetSPEnterpriseSearchPrimaryHostControllerSetSPEnterpriseSearchLinguisticComponentsStatusSetSPEnterpriseSearchContentEnrichmentConfigurationRemoveSPEnterpriseSearchContentEnrichmentConfigurationNewSPEnterpriseSearchContentEnrichmentConfigurationGetSPEnterpriseSearchLinguisticComponentsStatusGetSPEnterpriseSearchHostControllerSetSPEnterpriseSearchLinksDatabaseRepartitionSPEnterpriseSearchLinksDatabasesMoveSPEnterpriseSearchLinksDatabasesRemoveSPEnterpriseSearchTenantSchemaRemoveSPEnterpriseSearchTenantConfigurationRemoveSPEnterpriseSearchLinksDatabaseRemoveSPEnterpriseSearchFileFormatNewSPEnterpriseSearchLinksDatabaseNewSPEnterpriseSearchFileFormatNewSPEnterpriseSearchAnalyticsProcessingComponentImportSPEnterpriseSearchCustomExtractionDictionaryGetSPEnterpriseSearchLinksDatabaseGetSPEnterpriseSearchFileFormatSetSPEnterpriseSearchFileFormatStateGetSPEnterpriseSearchComponentGetSPEnterpriseSearchServiceApplicationBackupStore

    Yournewbestfriend

  • Scalable Architecture Examples

    SmallSearchFarm(~10MillionItems)|4PhysicalServers|6VirtualServers

  • Scalable Architecture Examples

    MediumSearchFarm(~40MillionItems)|8PhysicalServers|14VirtualServers

  • Scalable Architecture Examples

    LargeSearchFarm(~100MillionItems)|18PhysicalServers|28VirtualServers

  • Scalable Architecture Examples

    ???

    ExtremeSearchFarm(>100MillionItems)

  • ECM Impact

    Schemacanbemanagedbysiteadmins,reducingtheloadonsearchadministrator

    Schemaisconfiguredtoallowmoregranularity(query,retrieve,refine,sort,etc) Affectscontentindexsize

    Remoteresultsourcescanbecrawledlocallyandthenqueriedbyremotefarms.HugeimpactongeodistributedsearchThirdpartysolutionsmaybeabletofacilitateblendedsearchresults!

    Individualitemscanberecrawledeasily

    AutomaticURLbalancingincrawldatabasesminimizeshostnamerestrictionsforlargearchiverepositories

    Scalabilitylimitchangeswillhaveabigimpactonfarmdesignforlargearchivecontentrepositoriesinthenearfuture.

    Whatdoesitallmean?

  • Questions?

    StumptheChump(youprobablycan)

  • Thankyou!

    The end

    Search ArchitectureIntroductionUp for discussionA brief historyConcepts: SP2010 SP2013Concepts: New FunctionalityConcepts: New FunctionalityConcepts: New FunctionalityConcepts: New FunctionalityConcepts: New FunctionalityConcepts: New FunctionalityFAST Its in thereFAST Its in thereComponent Topology OverhaulComponent Topology OverhaulComponent Topology OverhaulComponent Topology OverhaulComponent Topology OverhaulComponent Topology OverhaulComponent Topology OverhaulSchema ManagementPowerShell Its unavoidablePowerShell Its unavoidableScalable Architecture ExamplesScalable Architecture ExamplesScalable Architecture ExamplesScalable Architecture ExamplesECM ImpactQuestions?The end


Recommended