FullText Indexing And Search Using Site Server And Exchange

1
Full-Text Indexing And Search Using Site Server
And ExchangeTom Rizzo ? thomriz_at_microsoft.com
5-402
2
(No Transcript)
3
This Session Covers

• Scenarios
• Overview
• Site Server search
• Exchange indexing overview
• Exchange deployment steps
• Developing a custom search solution
• Common errors and limitations
• Additional resources

4
Where To Get More Information

• SBN http//www.microsoft.com/workshop/server/sit
  eserver/siteexch/exchss.asp
• Site Server Update Web sitehttp//www.microsoft.
  com/siteserver/update
• Search Performance report
• Search deployment whitepaper (tips tricks)
• Database indexing MSDN article
• SDK update
• Evaluation guide

5
(No Transcript)
6
Scenarios

• Custom Helpdesk application
• Support site for internal IT apps
• Public folder for user questions
• FAQ auto-generated as questions get answered
• KnowledgeBase
• Repository for best practices
• Rich search allows people to find information
  easily
• Custom forms allow users to quickly enter data
  into application

7
Powerful, Multi-Store SearchMicrosoft Site Server

• Powerful
• Full text and property search
• Security integration for Microsoft Exchange and
  NTFS
• Site vocabulary and tagging
• Automatic language detection
• Extensive
• Single query across multiple stores
• HTTP, file system, Microsoft Exchange public
  folders, databases

8
Industrial Strength SolutionMicrosoft Site Server

• Scalable
• Distributed indexing
• Replicated query servers
• Incremental crawl reduces Net traffic
• Administration
• MMC, Web-based, command line, and scriptable
  administration

9
Exchange Indexing Features

• Index and search Public Folders
• Search properties, text, attachments
• Index and search custom properties
• Integrated with Exchange security
• Full and incremental crawl
• Search using web browser
• Read search result messages using OWA or Outlook

10
Search Architecture
11
Exchange Search Deployment Checklist
ü

• Capacity planning
• Configure Site Server accounts
• Configure Site Server for Exchange crawling
• Index and search Exchange properties

ü
ü
ü
12
Scalability And Performance

• Reference HW Dual p6/200, 128MB
• Little/no benefit from quad proc
• Limits
• 5M docs/catalog max, 32 catalogs max
• Crawl speed 10 messages/sec (30KB/sec)
• Query rate 11-15 Q/s (25-55Q/s without
  hitcount)
• Want physical memory gt property store
• Factors hit count, result set size, Exchange
  security
• Non-factors query complexity, catalog size (as
  long as property store fits in RAM)

13
Hit Count, Query Complexity
14
Result Count, Security
15
Catalog Size, Hitcount
16
Capacity Planning

• How many machines?
• Criteria crawl overnight (8 hrs), user queries
• 1 machine for up to 700k docs, 40k users
• Add machines for
• more documents
• more users
• fault tolerance
• How much RAM?
• Criteria Fit property store in physical RAM
• lt 150K docs 128MB RAM
• lt 1M docs 256MB RAM
• gt 1M docs 512MB

17
Configuration - Accounts

• Set Content Access account on Build Server
• Configure through Catalog Build Server Properties
  -gt Accounts
• Need Admin privilege on the Configuration node
  of Exchange Server
• Set Search service to run with same account
• Configure through Control Panel -gt Services
• By default, Site Server Search Service runs as
  SYSTEM
• Must have same privileges above
• Must have admin privileges on local machine

18
Configuration - Accounts

• Search query pages must have anonymous access
  turned off
• Diagnostic information can be found in
• Gatherer log
• Event viewer
• Query error return codes

19
ConfigurationExchange Settings

• Set Exchange Server configuration in Server
  properties (under Search MMC node)
• Exchange Server Name
• Outlook Web Access Server Name - Optional
• Exchange Server Site Name
• Exchange Server Organization Name
• If settings are incorrect, diagnostic information
  is in the gatherer log, event viewer and query
  error strings
• Must be set on the Catalog Build Server and
  Catalog Search Server if distributing indexes

20
What About Replication?

• If your PFs are replicated, Site Server can
  access your PFs as long as
• They are homed in the Site you specify for Site
  Server
• You use Public Folder Affinity for PFs in other
  sites
• Make sure Search has administrative access to all
  PFs youre crawling
• Increase timeout period if latency is a problem

21
Build And Search An Exchange Catalog

• Create a new Catalog Definition using wizard
• Do a full build
• Verify success by searching using the default
  search page

22
Exchange Properties

• Mapped to standard properties
• DocTitle (Subject)
• DocAuthor (From)
• FileWrite (SendDate)
• Description (300 chars from message)
• Additional Exchange properties
• MessageClassMessageDisplayName
  MessageDisplayCCMessageFolderName
• MessageFolderPath

23
Exchange Custom Properties

• Custom properties work the same way as HTML Meta
  Tags
• Available in the Meta property set
• Searchable by default (not retrievable)
• Text, date/time, and number supported
• Text properties can automatically be searched
• Use Meta_ltNamegt (e.g. Meta_myprop)
• For Custom Props with spaces use Meta_ltNamegt
• E.g., Meta_Total Expense

24
Custom Properties Dates And Numbers

• For date/time and number custom properties
• Edit DefineColumns.txt
• Default location c\Microsoft Site
  Server\Data\Search\Config
• Add entry corresponding to custom property in the
  Meta property set
• Use appropriate type
• Integer DBTYPE_I4
• Date VT_FILETIME
• Refer to documentation on Meta properties

25
Demo Building A Custom Search ASP Application

• Using the capabilities of Site Server, you can
  create rich, custom search applications
• These applications can search both standard props
  as well as your custom props in Exchange

26
Demo Accessing Searches

• Outlook Search techniques
• HTML Page self-hosted in Outlook
• Outlook Form driving HTML page
• Custom Outlook Forms
• DCOM connection to Site Server to enable native
  Outlook search

27
Specifying The Catalog
ltinput typehidden name"c6" value"_at_MessageFolder
"gt ltinput typehidden name"q6" size25
maxlength50 value"lt q6 gt"gt ltselect
name"ct" value"ct"gt ltoption value"Discussion,Pr
oductKnowledge,Business,list_Servers"gt All
Indexed Folders ltoption value"Discussion"gt
\Discussion
28
String Custom Properties
lttdgtltinput typehidden name"c7"
value"_at_META_Type"gt ltselect name"q7"gt
29
Date Properties
ltinput typehidden name"c3 value"_at_filewrite"gt
ltinput typehidden name"o3" value"gt"gt
ltselect name"q3"gt ltoption value"" ltif
Request("q3")"" then gtselectedltend ifgt
gt ltoption value"-1d" ltif q3"-1d" then
gtselectedltend ifgt gtday ltoption value"-1w"
ltif q3"-1w" then gtselectedltend ifgt
gtweek ltoption value"-1m" ltif q3"-1m" then
gtselectedltend ifgt gtmonth ltoption value"-1y"
ltif q3"-1y" then gtselectedltend ifgt gtyear
30
Returning Results OWA Or Outlook
if ExchangeViewer"both" or ExchangeViewer"outloo
k" then gt ltobject id"Exciol" height0 width0
CLASSID"CLSIDDAFD7A40-73FF-11D1-A811-00AA006EAC9
D" CODEBASE"/siteserver/knowledge/search/control
s/exciol.ocxversion5,5,2148,0"
TYPE"application/x-oleobject"gt lt/objectgt
31
Set Query Object
lt ' Set query object properties. set Q
Server.CreateObject("MSSearch.Query") '
Define all required query properties the query
itself, and all columns ' that will be used
in the results. Q.SetQueryFromURL(Request.Q
ueryString) Q.Catalog request("ct")
Q.OptimizeFor"nohitcount,performance"
Q.MaxRecords 10 Q.Columns "DocAuthor,
DocTitle, DocAddress, Description, Size,
FileWrite" gt
32
Create The Recordset
lt ' Create the recordset holding the search
results. on error resume next set RS
Q.CreateRecordSet("sequential") if err
ltgt 0 then createerror err.description
end if
33
Scroll Through The RS
lt ' Set up loop to iterate through results. Do
while not RS.EOF Response.Write
RS("Description") ' Increment the results.
RS.MoveNext RecordNum RecordNum 1
Loop gt
34
Common Errors

• Specifying the wrong Site or Organization
• The crawling/searching account must be in the
  administrators group on Site Server computer
• Content access account must have Admin rights
  on Exchange Server Configuration node
• Search Service must be manually configured to run
  as a user with admin privilege, not SYSTEM
  (which is the default)

35
Troubleshooting Failed Crawls

• Start with Event Viewer (Application log)
• Next, look at Gatherer Log (from MMC)
• Logs per item failures by default
• Can turn on success logging and rules logging
• Confirm network connectivity to content
• Ensure that the connectivity is checked for the
  default content access accounts
• Validate permissions of content access account

36
Other Sessions

• Noam Topazs Notes session later today! Hell
  have CDs with source for the COM Add-in!

37
Questions?
38
(No Transcript)