database-disaster-recovery

Download Report

Transcript database-disaster-recovery

Choosing Disaster Recovery Solution
for Database Systems
EECS711 : Security Management and Audit
Spring 2010
Presenter : Amit Dandekar
Instructor : Dr. Hossein Saiedian
1
Contents
•
•
•
•
•
Database failures types
Availability solutions
Availability mechanisms
Recovery procedures
Conclusion
2
Failure types
• Database failure types
– Transient
– Crash
– Media
– Site
– Operator
– Malicious
Least Severe
Most Severe
3
Failure types
Others
Natural disaster
3%
6%
30%
Power failure
Hardware related
16%
DBMS related
20%
25%
Source: Forrester Research, Inc.
4
Human error
Availability solutions
• Two categories of availability solutions
– Sporadic un-availability of database
• Mission critical systems
• Online transaction processing systems
– Complete un-availability of database
• Data warehouse
• Decision support systems
5
Availability solutions
• Protect against sporadic unavailability
• Used to guard against sporadic outages
• Implementation may be co-located
– Geographically distributed to protect against site failure
•
•
•
•
Recovery time is expected to be within minutes or less
Recovery point is within minutes or immediate
More complicated to deploy and expensive
Expensive
6
Availability solutions
• Protect against complete unavailability
•
•
•
•
•
•
Used to guard against disasters
Geographically distributed implementation
Recovery times within hours or days
Recovery point may be within hours or days
Relatively less complicated to deploy
Less expensive
7
Availability mechanisms
• Data synchronization
– Online synchronization
• Primary and secondary are always synchronized
– Allows immediate primary takeover
– Comes with communication and performance overhead
Offline synchronization
– Offline synchronization
• Synchronization performed when no active
transactions occurring
• Typically backup site synced periodically
– May lose updates in case of disaster
8
Availability mechanisms
• Data replication
– Active replication
• Data is transferred and processed
• Can share workload with primary site
– Secondary should have enough processing power
– Passive replication
• Data is transferred and stored without processing
• Typically use Redundant Array of Disks (RAID)
– Guards against media failure
• Remote mirror required to recover from site failure
9
Popular availability solutions
Others
5%
Mirroring and
replication
20%
40%
Log shipping
35%
Source: Forrester Research, Inc.
10
Backup tapes
Disaster recovery procedure
• Recovery when using active replication or
online synchronization
– Hot site is made the primary site by system admin
• Automated fail-over may treat transient failures as
disasters
– Recovery time can be as short as few minutes
11
Disaster recovery procedure
• Three common disaster recovery approaches
when using off-line or passive mechanisms
– Sledgehammer
• Rebuild entire database from scratch
• Off-line approach for non-critical, non-volatile data
– Behind the back copying
• Copy and rebuild table spaces and index datasets
– Scalpel
• Performs restoration of data at granular level
• Restore one table at a time in order of priority
12
Conclusion
• Evaluate and identify your database
availability requirements
– Protect against sporadic un-availability
– Protect against complete unavailability
• Choose appropriate availability mechanism as
disaster recovery technique
– Active replication or online synchronization
– Passive replication or offline synchronization
13
References
Choy, Manhoi, Hong Va Leong, and Man Hon Wong. "Disaster recovery techniques for
database systems." Commun. ACM 43.11es (2000): 272-280.
Yuhanna, Noel. "ForrTel: Making Your Enterprise Database Highly Available." Forrester
Research. Forrester Research, Web. 19 Feb. 2010.
<www.forrester.com/Events/Content/0,5180,1131,00.ppt>
14