Tuesday, June 20, 2006

Thanks to Jason Medero - TechEd Brain Dump #2 Moss 2007 Advanced-Capacity-Planning-Infrastructure-Deployment

TechEd Brain Dump #2 Moss 2007 Advanced-Capacity-Planning-Infrastructure-Deployment



Here are a few notes that I took from the Joel Oleson and John Nisi’s presentation on Moss 2007 Infrastructure topics and admin architecture.  Some good stuff I hope everyone can read my somewhat disorganized note taking!


 


 



  • Shared services


    • Grouped, high-value, resource intensive services

    • One to many per farm

    • Inter-farm capable

 



  • NO MORE TOPOLOGY RESTRICTIONS!!


    • Factors to consider:


      • Data composition

      • User load

      • Long-running operations

      • Performance

      • Availability & reliability

      • Network considerations

 


          Beta 2 Hardware Requirements:  (Will change come RTM)


o   Single box installation


§  CPU:  2.5 GHz


§  Memory:  2 GB recommended, 1 GB minimum


§  HDD:  Scenario dependent


o   Farm Deployment


§  Web server: 2.5 GHz, 2 GB RAM


§  App server: Dual proc 2.5 GHz, 2 GB RAM


§  SQL: Dual proc 2.5 GHz; 2gb RAM


§  Support both 32 & 64-bit






  • MOSS 2007 Disaster Recovery


    • Backup and Restore methods


      • 2-Stage Recycle Bin for documents and lists

      • Site-level backup/restore via STSADM

      • Integrated backup/restore UI for web application and farm

      • VSS writer for farm backup

      • SQL Server backup/restore

    • Mirror/failover farm


      • Replicate primary farm on secondary system

      • SQL log shipping transfers content DB data

      • Must manually replicate configuration changes

      • On disaster, router switches traffic in minutes

 



  • Servers have roles


    • WFE

    • App server:  Indexing, Search, Excel Calc, Project

    • Db server


  • You can create a farm of any size of each Server Role

  • Some guidelines:


    • At least 1 server specified as Index

    • No more than 8 WFE for each SQL Server

 



  • Watch-Outs for Shared Services


    • Farm:  SSP web app pool account CANNOT be Network Service (local accounts)

    • 1 SSP admin site allowed in a single web application ---->  \admin\ssp

    • Central admin operators == SSP site administrators

    • Closely manage security when switching associations

 



  • Use cases for multiple SSPs (Shared Service Providers)


    • Secure isolation of services and service data


      • Hosted environments

      • Restricted sites

      • Organizational / Political concerns

 



  • Special Rights


    • Central Admin SPADMIN must be LocalSystem on all servers


      • DBO for all DBs

      • DB Creator

      • SQL Security Admin

    • Shared Services :


      • DBO for content DB

      • R/W to SSP DBs

      • Read from config DB

      • R/W to SSP DBs

      • R/W to content DBs

      • Read from config DB

 


o    Site Rights


          DBO for content DB


          R/W to SSP DBs


          Read from config DB


 



  • Security Best Practices


    • Unique accounts for the following:


      • Farm account

      • SSP process account


        • NOTE: Cannot be Network Service in a farm config.

      • SSP shared web service account

      • Content app pool

    • Kerberos on (default = NTLM)


      • Each process account must be a registered SPN to work

    • SSL enabled (default = off*)


      • Turn on for admin sites & server to server

      • Warning provided on credentials pages if SSL is off

    • SPAdmin service:


      • Single server: Off (recommend â€�On’ for OSS)

      • Farm: On

 



  • Switch over to Kerberos for quicker authentication highly recommended to switch from NTLM and provide a SPN (Service Principal Name) for a more secure environment.

 



  • Additional configuration


    • Web load balancing (incl. static IP)

    • SSL/Kerberos

    • ASP. Net pluggable authN provider

    • Caching – for performance

 


 


Cheers mates!

0 Comments:

Post a Comment

<< Home