There are many different disaster recovery and high availability options for SQL Server. Making decisions on the most effective DR & HA strategies can be complex, especially when you throw into the mix various SAN and network topologies.
This presentation is focused on the management and operational decisions that are made when planning DR and HA for production SQL Server environments. It is targeted towards Senior DBAs, CIO, IT Manager, database services managers.
Topics include:
Log Shipping
Database Mirroring
Always On High Availability groups
Replication
Clustering
Licensing
2. Disclaimer
We recommend that you seek further professional advice before
deciding on the suitability of any recommendations in this presentation
for you. While all care is taken to ensure information is accurate, we do
not make any guarantees about the suitability of advice and advice
given may contain technical inaccuracies or topographical errors. The
answers provided are our own opinions and may differ from advice
provided by Microsoft and/or other SQL Server professionals. In no
event shall we be liable for any special, incidental, indirect, economic or
consequential damages or for loss of profit, revenue or data howsoever
caused, regardless of whether we could foresee or was advised of the
possibility or likelihood of such loss or damage.
3. Disaster Recovery & High Availability Technologies (SQL
Server 2008, 2008 R2 & 2012 Standard & Enterprise)
Contents
Log Shipping
Replication
Clustering
Database Mirroring
Always On High Availability groups
Leveraging SQL Server 2012 core licensing for High Availability /
DR implementations
4. Disaster Recovery & High Availability Technologies
What’s the difference between Disaster Recovery & High
Availability?
DR
Log Shipping
Async database mirroring
Async AlwaysOn Availability Groups
HA
Clustering
Sync database mirroring
Sync AlwaysOn Availability Groups
Replication
5. DR - Log Shipping
Performed at database level
Logged operations copied from a primary to secondary offsite
database
Consists of 3 operations
• Backup TL at source
• Copy of TL from source to destination
• Restore TL at destination
Prerequisites
• Recovery model of FULL/BULK-LOGGED
• Create network share for transaction log backups
• SQL Server Agent must be running
6. DR - Log Shipping continued ….
Generally, failover is manual process
Configuration options determine data loss
• Backup/restore schedule
Other options
• Compress backups?
• How long you want to keep backup files?
• Secondary db RECOVERING or STANDBY state?
8. Replication
Performed at database level
Distribute & replicate data automatically
Integrating data from multiple sources
Data warehousing & reporting
Typical topology - Publisher, Distributor & Subscriber
One-way or bidirectional
depending upon type
9. Replication continued ..
Options
Publication type: Snapshot, Transactional & Merge
• Advantages of replication
• Multiple sites have copies of data
• Separate applications for read/write operations
• Increased performance
• Disadvantages of replication
• Can be complicated to manage changes
• Multiple SQL licenses
10. HA - Clustering
Performed at an instance level
One virtual server name, multiple nodes
Shared storage
Best Practices
• Server dedicated to SQL Server
• Identical servers
• Use static IP address
• Plan how to deal with risks - SAN failure & DR options
11. HA – Clustering continued …
Advantages
• Patching SQL Server is easier
• Eliminate downtime
• Logins, jobs, system db’s all failover
• Don’t license passive node
Disadvantages
• Is generally not a DR solution
• Single copy of db
• Shared storage issue
• Expensive – x 2 servers and shared storage
12. DR & HA - Database Mirroring
Performed at a database level
Principle, Mirror & optional Witness
Cannot mirror master, msdb, tempdb or model
Apply stream of database log records to db mirror
Used for DR (async) & HA (sync)
Prerequisites
• FULL RECOVERY
• Principal & Mirror require same edition of SQL Server
Options
• Supports automatic failover (witness)
• Sync OR async
13. DR & HA - Database Mirroring continued …
Witness Database
• It’s role is to determine availability between Principal & Mirror
• Used for HA not DR
• Ideally, located in a third datacentre
Advantages
• Since SQL Server 2008 -
• The principle compresses the transactions before sending
to mirror.
• Automatically detects & attempts to recover storage
corruption
Limitations
• Can’t read mirror – unless using snapshot
• Can only fail over one database at a time
• Doesn’t copy SQL logins or jobs
15. AlwaysOn Availability Groups
SQL Server 2012
Next evolution of db mirroring
HA+DR solution
Windows Server Failover Cluster
license Sync – auto license
Sync – auto
failover Async – failover
Async – forced
forced
SQL2012PROD1 SQL2012PROD2 SQL2012DR1
Primary Replica Secondary Replica Secondary Replica
Read-Write Disallow connections Disallow connections
<---------RockSolid Availability Group----->
Melbourne Primary data centre Sydney DR data centre
16. AlwaysOn Availability Groups continued ..
Prerequisites
• Windows Server Failover Clustering (WSFC)
• SQL Server 2012 Enterprise license
• Same instance collation
• Advise same drive letters
Options
• Failover groups contain multiple db’s
• Supports up to 5 availability groups
• Supports async & sync failover
17. AlwaysOn Availability Groups continued ..
Replica Mode – Automatic failover (sync - auto) OR High
Safety (sync - manual) OR High Performance (async - manual)
Connection Mode in Secondary Role – Disallow connections OR
Allow only read-intent connections
18. AlwaysOn Availability Groups continued ..
Advantages
• Recovery of corrupt pages are automatically attempted
• HA & DR solution
• Replicas can be used for reporting & backup operations
Disadvantages
• Enterprise feature
• No delay in applying transaction log updates if you want that
control
19. Leveraging SQL Server licensing for High Availability / DR
implementations
SQL Server customers can run one supportive passive instance
• The passive server must be truly passive
• One secondary passive server is ONLY allowed
• When licensing Per Core you must license the highest number
of core licenses even if this is the passive node
20. Leveraging SQL Server licensing for High Availability / DR
implementations
2008 2008 2008 R2 2008 R2 2012 2012
Enterprise Standard Enterprise Standard Enterprise Standard
Log Yes Yes Yes Yes Yes Yes
Shipping
Database Yes Yes sync – Yes Yes sync – Yes Yes sync –
Mirroring full safety full safety full safety
only only only
Always on n/a n/a n/a n/a Yes – No
High requires
Availability Windows
clustering
Clustering Yes Yes – 2 Yes Yes – 2 Yes Yes – 2
node node node
support support support
21. Leveraging SQL Server licensing for High Availability / DR
implementations
Server 1 – 4x core licenses required Server 2 –
ACTIVE PASSIVE
4 x core VM 4 x core VM
Server 2 –
PASSIVE
Server 1 – 4 x core VM
ACTIVE
4 x core VM Server 2 –
8x core licenses required PASSIVE
4 x core VM
22. Leveraging SQL Server licensing for High Availability / DR
implementations
Server 2 –
PASSIVE
Server 1 – 4 x core VM
ACTIVE
4 x core VM
Server 2 –
12x core licenses required PASSIVE
8 x core VM
8x licenses required as
license requirement based
on passive server