High Availability Solutions

1
High AvailabilitySolutions

• SQL 2005
• Database Mirroring and Log Shipping
• SQL Clustering

2
Database Mirroring

• Implemented on a per database level
• Transaction logs sent from principle to mirror
• Provides a warm standby in case of failure
• Principal and mirror must be separate SQL 2005
  SP1 servers (Enterprise or Standard)
• Optional witness server to monitor primary and
  mirror to ensure both are working
• Automatic failover - on failure the witness will
  turn the mirror into the principle
• WSS v3 Whitepaper coming soon

3
Database Mirroring

• Three modes that can be used
• High Availability synchronizes transaction
  writes on both servers and enables automated
  failover (requires a witness server).
• High Protection synchronizes transaction writes
  on both servers, manual failover.
• High Performance writes are not synchronized on
  both servers, offering some performance gains.
  Assumes everything will complete successfully on
  the mirror.

4
Database Mirroring
Supported Topology
Unsupported Topology
Supported Topology
5
Database Mirroring
Witness Server
Everything ok?
Everything ok?
Encrypted channel
Principal
Mirror
6
SharePoint and Mirroring

• When failover occurs need to notify SharePoint
• Need to tell these DBs that a new server is now
  the principal
• Configuration Database, Admin Content Database
• Content Database
• Shared Service Provider Database (MOSS)
• Search Database (MOSS)
• Content DBs attach reattach
• Config DB and admin content DB
• stsadm.exe -o renameserver -oldservername
  ltoldServerNamegt -newservername ltnewServerNamegt
• Etc. (look for whitepaper)

7
Log-Shipping

• Backup/Restore based technology that relies on
  transaction log files
• Need to have a shared folder that contains the
  log file backups
• Configure the frequency of backups and shipping
• No automatic failover
• Allows you to replicate data to several databases
  (one to many)

8
Log-Shipping Mirror Farm
I want the minimal delay when my data center
goes down! I need another farm to test my
solutions before I put them into production!

• Mirror / failover farm
• Warm backup for quick content recovery
• Replicate primary farm on secondary system
• SQL log shipping transfers content DB data
• On disaster, router switches traffic in minutes
• Limitations
• Must manually replicate configuration changes
• Errors will be log shipped to mirror farm
• Requires additional hardware
• Search indexer must run separately on mirror farm
• Secondary farm must be read-only

9
Log-shipping Mirror Farm
10
SQL Failover Clustering
If my SQL server blows up (hardware failure),
how do I keep my SharePoint farm running?

• Hot backup short recovery time
• Unattended application availability automatic
  failover capability without administrator
  intervention

SharePoint Web Front Ends
Server Cluster
Heartbeat
Node B
Node A
Shared Disk Array
11
Log-Shipping vs. Clustering