How Oracle Database 10g Revolutionizes Availability and Enables the Grid

1
(No Transcript)
2
How Oracle Database 10g Revolutionizes
Availability and Enables the Grid
Session id 40164

• Juan LoaizaVice President, Systems Technologies
• Oracle Corporation

3
From High Quality Parts to High Quality Systems

• Traditionally Low Cost Low Quality
• High quality systems were built by combining high
  quality, high cost parts Mainframe model
• Oracle enables a new model
• Oracle combines high volume inexpensive
  processors and storage to produce a high quality
  system
• Unbreakable Inexpensive Systems

4
Low Cost Fault Tolerance
Computer Failures
Grid Clusters Low Cost Fault Tolerance
UnplannedDowntime
Data Failures
System Changes
PlannedDowntime
Data Changes
5
Commercial Grids and Availability

• Grid pools standard low cost nodes and modular
  disk arrays
• Perfect for RAC HA
• Failover can happen to any node on the grid
• Grid load balancing will redistribute load over
  time

Designed to Tolerate Failures
6
New in 10g for Availability

• Integrated clusterware
• Integrated, less moving parts, better tested
• Avoid operator errors by reducing cross product
  coordination
• Built in integration with Application Server for
  failing over connection pools
• Faster failover between nodes
• Single digit seconds

7
New Economics forData Protection Recovery
Computer Failures
Disk Based Recovery
UnplannedDowntime
Data Failures
Trade cheap disk space for expensive downtime
System Changes
PlannedDowntime
Data Changes
8
New World Disk Based Data Recovery

• Disk economics are close to tape
• Disk is better than tape
• Random access to any data
• We rearchitected our recovery strategy to take
  advantage of these economics
• Random access allows us to backup and recover
  just the changes to the database
• Backup and Recovery goes from hours to minutes

1980s - 200 MB
1000x increase
2000s - 200 GB
9
Resiliency using Low Cost Storage
Storage Failure
Computer Failures
UnplannedDowntime
Human Error
Data Failures
Corruption
System Changes
PlannedDowntime
Site Failure
Data Changes
Four Failure Types
10
Data Mirroring with ASM

• ASM mirrors data across inexpensive modular
  storage arrays
• No additional logging or expensive NVRAM to
  recover mirrors
• Database logging recovers mirrors
• Automatically remirrors when disk or array fails
• Designed to tolerate failures

Failure Resiliency using Low Cost Storage
11
Collapsing the Cost of Human Error
Storage Failure
Computer Failures
UnplannedDowntime
Human Error
Data Failures
Corruption
System Changes
PlannedDowntime
Site Failure
Data Changes
12
Human Error
Single Biggest Cause of Downtime

• Goal is to quickly analyze and repair
• For Localized damage
• Need surgical analysis and repair
• Example deleted wrong order
• For Widespread damage
• Need complete back-out to avoid long downtime
• Example batch job deletes this months orders

13
Flashback Time Navigation

• Flashback Query
• Query all data at point in time
• Flashback Versions Query
• See all versions of a row between two times
• See transactions that changed the row
• Flashback Transaction Query
• See all changes made by a transaction

Select from Emp AS OF 200 P.M. where
Tx 3
Select from Emp VERSIONS BETWEEN 200 PM and
300 PM where
Tx 2
Tx 1
Select from DBA_TRANSACTION_QUERY where xid
000200030000002D
14
Flashback Database

• A new strategy for point in time recovery
• Flashback Log captures old versions of changed
  blocks
• Think of it as a continuous backup
• Replay log to restore DB to time
• Restores just changed blocks
• Its fast - recover in minutes, not hours
• Its easy - single command restore
• Flashback Database to 205 PM

Disk Write
New Block Version
Old Block Version
Rewind button for the Database
Data Files
FlashbackLog
15
Flashback Error Correction

• Recovery at all levels
• Database Level
• Flashback Database restores the whole database to
  time
• Uses Flashback Logs
• Table Level
• Flashback Table restores rows in a set of tables
  to time
• Uses UNDO in database
• Flashback Drop restores a dropped table or a
  index
• Recycle bin for DROPs
• Row Level
• Restore individual rows
• Uses Flashback Query

Database
Customer
Order
16
Flashback for All Users

• END USER
• Flashback Query
• Flashback Versions Query

• DEVELOPER
• Flashback Versions Query
• Flashback Transaction Query
• Flashback Table

• DATABASE ADMIN
• Flashback Database
• Flashback Drop

• SYSTEM ADMIN
• Data Guard

17
Revolution in Recovery

• Flashback Revolutionizes Recovery
• Operates on just the changed data
• Time to correct error equals time to make error
• Minutes instead of hours
• Flashback is Easy
• Single command instead of complex procedure

Correction Time Error Time f(DB_SIZE)
18
Prevention Recovery of Corruptions
Storage Failure
Computer Failures
UnplannedDowntime
Human Error
Data Failures
Corruption
System Changes
PlannedDowntime
Site Failure
Data Changes
19
Oracle End-to-end Data Validation
Blocks validated and Protection info added to
block
Oracle

• H.A.R.D. Hardware Assisted Resilient Data
• Prevents corruption introduced in IO path between
  DB and storage
• Initially introduced in Oracle9iR2
• 10g HARD provides
• Better checks
• All file types block sizes checked
• DB, log, archive, backup, etc.
• A.S.M. enables HARD without using RAW devices
• Supported by major storage vendors

A.S.M.
Volume Manager
Operating System
Host Bus Adapter
SAN Virtualization
SAN Interface
Storage Device
Protection info validated by storage device
20
Flash Recovery Area

• Fully automatic disk based backup and recovery
• Set and Forget
• Nightly incremental backup rolls forward recovery
  area backup
• Changed blocks are tracked in production DB
• Full scan is never needed
• Dramatically faster (20x)
• Blocks validated to prevent corruption of backup
  copy
• Use low cost ATA disk array for recovery area

Flash RecoveryArea
Nightly Apply Validated Incremental
Weekly Archive To Tape
DatabaseArea
Two Independent Disk Systems
21
Low Cost No Compromise Disaster Recovery
Storage Failure
Computer Failures
UnplannedDowntime
Human Error
Data Failures
Corruption
System Changes
PlannedDowntime
Site Failure
Data Changes
22
Existing Site Recovery Tradeoffs
Production Database
Standby Database
Transaction Shipping
Reporting On DelayedData
4 Hour Delay Apply

• User can delay log apply to protect from user
  errors but
• Failover takes hours
• Reports run on hours old data
• After failing over to standby, production DB must
  be rebuilt
• Production has updates that did not get to
  standby

23
Low Cost No Compromise Disaster Recovery
Reporting On Real TimeData
Standby Database
Transaction Shipping (Real Time Apply)
Production Database
Some Nodes Used for Other Computing
No Delay
Flashback Log
Flashback Log

• Flashback DB removes need to delay apply of logs
  to correct errors
• Flashback DB removes the need to reinstantiate
  primary on failover
• Real-time log apply enables real-time reporting
  on standby
• Data Guard works transparently across GRID
  clusters
• Standby can use fewer CPU resources than primary

24
Highest Data Protection Lowest Cost
Dramatic Advances in Ease of Use
Data Guard Site Failure Protection
Combine the Features to Achieve Any Level of
Data Protection
25
Other Protection Enhancements

• Compression of archive logs and backups
• Automated failover to a previous backup when
  restore discovers a missing or corrupt backup
• Automated recovery through a previous point in
  time recovery - recovery through resetlogs
• Automated creation of new files during recovery
• Automated channel failover on backup or restore
• Automated tablespace point-in-time recovery
• Full DB begin backup command for faster mirror
  split
• Improved Recovery Parallelism (2 to 4 X)
• Tablespace Rename
• Proxy (third-party) Backup for archive logs
• Time window based throttling of backups

26
Other 10g Data Guard Enhancements

• SQL Apply Enhancements
• Support for Longs
• Support for multi-byte CLOBs and NCLOBs
• Support for Index Organized Tables without
  overflow or LOB segments
• Instantiation of logical standby with no quiesce
  of primary
• Generic Data Guard Enhancements
• Data Guard Broker support for RAC

27
No Cost System Changes
Goal
Computer Failures

• Allow any change to the system with no downtime

UnplannedDowntime
Data Failures
Online Reconfiguration
System Changes
PlannedDowntime
Rolling Upgrades
Data Changes
28
No Cost System Changes Capacity on Demand

• CPU
• Add/remove CPUs on SMP online
• Cluster Nodes
• Add/remove cluster nodes online
• No data movement needed
• Memory
• Grow and shrink shared memory and buffer cache
  online
• Auto tuning of memory online
• Disk
• Add/remove disks online
• Automatically rebalance
• Move datafiles

29
Rolling Patch Upgrade using RAC
Oracle Patch Upgrades
Clients
Clients
Patch
A
B
B
A
B
1
2
Operating System Upgrades
Initial RAC Configuration
Clients on A, Patch B
Patch
A
A
B
Hardware Upgrades
3
4
Upgrade Complete
Clients on B, Patch A
30
Rolling Release Upgrade using Data Guard
Logs Ship
Logs Queue
Upgrade
Patch Set Upgrades
Clients
Clients
1
2
Major Release Upgrades
Version X
Version X
Initial SQL Apply Config
Upgrade node B to X1
Logs Ship
Logs Ship
Cluster Software Hardware Upgrades
Clients
Clients
3
4
X1
Run mixed to test
Switch to B, upgrade A
31
No Cost Data Changes
Goal
Computer Failures

• Competitive pressures demand continual change
• Need to change data with no interruption to the
  application
• location, format, indexing, or even definition

UnplannedDowntime
Data Failures
System Changes
PlannedDowntime
Online Redefinition Evolution without Interruption
Data Changes
32
Online Redefinition

• All indexing operations can be done online
• Create new index, move index, defragment index
• Tables can be Reorganized Redefined online
• Table contents are copied to a new table
• Defragments and allows changing location, table
  type, partitioning
• Contents can be transformed as they are copied
• Can change columns, types, sizes - specified
  using SQL Select

Transform
Copy Table
Source Table
Result Table
GUI interface to make it Simple
Store Updates
Update Tracking
Continuous Queries Updates
Transform Updates
33
Online Redefinition Enhancements

• Enhanced Online Table Redefinition
• Easy cloning of indexes, grants, constraints,
  etc.
• Convert from long to LOB online
• Allow unique index instead of primary key
• Change tables without recompiling stored
  procedures
• Stored procedures can depend on the signature of
  a table instead of the table itself
• Online Segment Shrink
• Return unused space within the blocks of a
  segment to the tablespace

34
Maximum Availability Architecture (MAA)

• Operational Practices are key
• Technology alone is not enough
• MAA is a blueprint for achieving HA DR
• Tested, validated, and documented best practices
• Database, Storage, Cluster, Network
• 10 person year effort
• otn.oracle.com/deploy/availability

M.A.A. How to Prevent, Tolerate, Recover
From Outages
Maximum Availability Unbreakable Architecture
Best Practices
35
Highest Availability at Lowest Cost

• Highest Availability
• Fault Tolerant Clusters
• Flashback Error Correction
• Automated Disk Backup
• No Compromise Disaster Recovery
• Rolling Upgrades
• Online Redefinition
• At Lowest Cost
• Low Cost Grid servers
• Low Cost Modular Storage Arrays
• Automated Simple to Use

Oracle10g is Unbreakable Inexpensive
36
Next StepsHigh Availability Sessions from Oracle
Wednesday in Moscone Room 304
Tuesday in Moscone Room 304

• 1100 AM
• How Oracle Database 10g Revolutionizes
  Availability and Enables the Grid
• 330 PM
• Oracle Recovery Manager (RMAN) 10g Reloaded
• 500 PM
• Proven Techniques for Maximizing Availability

• 830 AM
• Oracle Database 10g - RMAN and ATA Storage in
  Action
•  
• 1100 AM
• Oracle Data Guard Maximum Data Protection at
  Minimum Cost
•  
• 100 PM
• Oracle Database 10g Time Navigation Human-Error
  Correction
•  
• 430 PM
• Data Guard SQL Apply Back to the Future

For More Info On Oracle HA Go To
http//otn.oracle.com/deploy/availability/
37
Next StepsHigh Availability Sessions from Oracle
Thursday
830 AM in Moscone Room 304 Oracle Database 10g
Data Warehouse Backup and Recovery Automatic,
Simple, Reliable 830 AM in Moscone Room
104 Building RAC Clusters over InfiniBand
For More Info On Oracle HA Go To
http//otn.oracle.com/deploy/availability/
38
Q

A
Q U E S T I O N S
A N S W E R S
39
New Oracle Database 10g HA Features

• Clusters
• Portable Clusterware
• Cluster file system for Linux Windows
• Automated Patching
• Data Guard SQL Apply
• Support for Longs
• Support for multi-byte CLOBs and NCLOBs
• Support for Index Organized Tables
• Simplified zero data loss failover
• Real time apply allows real time reporting
• Zero downtime instantiation

• Rolling Upgrades
• Rolling Upgrades Using Data Guard SQL Apply
• Online Redefinition
• Support of Unique Indexes
• One Step Cloning of Dependent Objects
• Columns can be Populated Using Sequences
  Sysdate
• Signature Based Dependency Tracking Using
  Synonyms
• Online Segment Shrink

• Data Guard Generic
• Data Guard Broker support for RAC
• Named Data Guard Configurations
• Real Time Apply
• Flashback Standby Database
• Flashback Reinstantiation
• Improved Recovery Parallelism

40
New Oracle Database 10g HA Features

• Flash Backup Recovery
• Automated Management of BR Disk Space
• Simplified Backup Using Image Copy
• Change Aware Incremental Backups
• Incrementally Updated Backups
• Compressed archive logs
• Tuning
• Improved Recovery Parallelism
• Faster Instance Startup Cache Warm

• Flashback
• Flashback Drop
• Flashback Row History
• Flashback Table
• Flashback Transaction History
• Flashback Database
• Better map of time to SCN for flashback query
• LogMiner
• Automated Specification of Logs to Mine
• Support for Shared Server Configurations
• Fine Grained Supplemental Logging

• Backup Recovery
• Simplified Recovery Through Resetlogs
• Restore Tolerates Missing Backups
• Proxy Backup of Archives
• Automated TSPITR Instantiation
• Full DB Begin Backup
• Automated Backup Channel Failover
• Simplified RMAN cataloging of backup files
• Automated File Creation during Recovery
• Drop Database
• Rename Tablespace

41
(No Transcript)