1 <Insert Picture Here> Next-Generation Oracle Database Availability – A Sneak Preview Juan Loaiza Senior Vice President, Oracle Corporation The following is intended to outline our general product direction. It is intended for information purposes only, and may not be incorporated into any contract. It is not a commitment to deliver any material, code, or functionality, and should not be relied upon in making purchasing decisions. The development, release, and timing of any features or functionality described for Oracle’s products remain at the sole discretion of Oracle. 3 Availability Focus Area • Oracle database development launched the “Database Focus Area” in 1998 (8i release) • Database availability was primarily provided by underlying hardware/OS platform • Goal was to divorce the database from the limitations of the hardware platform • Pool and virtualize underlying hardware resources • Create fault-tolerance using low cost hardware • Create innovative solutions to long standing availability problems • e.g. Human error, corruptions, etc. Establish a New Standard for High Availability 4 Old Standard – HA Components at High Cost Idle Cluster Failover Server Single SMP Server Premium Priced Idle DR Systems Clusterware & Volume Manager No Good Solution for: • Human Errors • Data Changes Premium Priced Single • Software Upgrades DB Backup Module Storage Array • Corruption Protection 5 New Standard – Oracle Maximum Availability Architecture • Best availability solution • For every outage category • At lowest cost • Low cost grid servers and storage • No idle hardware • Bundled clusterware, volume manager, and DR • Integrated, automated, and simple to use Best Availability AND Lowest Cost • High availability for the masses 6 Best Computer Failure Protection - at Lowest Cost Computer Failures Real Application Unplanned Clusters Downtime Data Failures System Planned Changes Downtime Data Changes 7 RAC Availability • RAC pools standard low cost servers • Great for Scalability • Perfect for Availability Database • No Idle Resources • Failover to any other node • Integrated clusterware Storage • Application notification • Thousands of production customers Designed to Tolerate Failure of Low Cost Servers 8 Next Release RAC Sneak Preview • Oracle focus is on maturity and stability • Fine tuning performance, scaling, failover, management • Competitors have still failed to deliver an OLTP cluster solution • Faster and more robust handling of instance crashes and hangs • Most of this also in 10.2.0.3 • Transaction tagging • Quickly identify which transactions committed before failure • Seamless integration with XA and Microsoft Transaction Server • Optimized Cache Fusion protocols 9 Best Data Protection - at Lowest Cost Computer Failures Unplanned Continuous Data Downtime Data Protection Failures Trade cheap disk space for expensive downtime System Planned Changes Downtime Data Changes 10 Best Storage Failure Protection - at Lowest Cost Computer Storage Failure Failures Unplanned Human Error Downtime Data Failures Corruption System Planned Changes Site Failure Downtime Data Changes Four Data Failure Types 11 Data Mirroring with ASM • ASM mirrors data across low cost modular storage arrays • Automatically remirrors when disk or array fails Database • ASM is free • Next Release Sneak Preview • Automatically repair corrupt blocks Storage from mirror copy • Temporarily freeze I/O to unresponsive storage and replay writes when it returns Designed to Tolerate Failure of Low Cost Storage 12 Best Human Error Protection - at Lowest Cost Computer Storage Failure Failures Unplanned Downtime Human Error Data Failures Corruption System Planned Changes Site Failure Downtime Data Changes 13 Investigate Errors using Time Navigation • Flashback Query • Query all data at point in time select * from Emp AS OF ‘2:00 P.M.’ where … Flashback Version Query – See all versions of a row between times – See transactions that changed the row Tx 3 select * from Emp VERSIONS BETWEEN ‘2:00 PM’ and ‘3:00 PM’ where … Tx 2 Flashback Transaction Query – See all changes made by a transaction Tx 1 select * from FLASHBACK_TRANSACTION_QUERY where xid = ‘000200030000002D’; 15 Flashback Error Correction Database • Correct errors at any level • Database Customer • Flashback Database - restore database to time • Table • Flashback Table - restore contents of tables to time • Flashback Drop - restore Order dropped table • Row • Flashback Query - restore individual rows 17 Revolution in Recovery Traditional Recovery 80 70 60 • Flashback Revolutionizes Error Recovery 50 40 • Operates on just changed data 30 • Time to correct error equals time to make error 20 10 Flashback • Minutes instead of hours 0 Time To Recover (minutes) Correction Time = Error Time + f(DB_SIZE) • Flashback is Easy • Single command instead of complex procedure • Very low performance overhead – less than 2% 19 Sneak Preview of Next Release Flashback 20 Next Release Flashback • Flashback Transaction • Push button back-out of a transaction and all conflicting transactions • Optimized Flashback Database • Lower overhead for inserts and direct loads 21 Flashback Data Archive Select * from orders • Long term retention - years AS OF ‘Midnight 31-Dec-2004’ • Automatically stores all changes to selected tables in Flashback Data Archive • Archive cannot be modified ORDERS • Old data purged per retention Archive policy Tables • View table contents as of any time using Flashback SQL • Uses User Flashback • Change Tracking Tablespaces Data Archive • ILM • Long term history Oracle Database • Auditing • Compliance Total Recall 22 Best Corruption Protection - at Lowest Cost Computer Storage Failure Failures Unplanned Human Error Downtime Data Failures Corruption System Planned Changes Site Failure Downtime Data Changes 23 RMAN - Best Tool for Repairing Corruptions • RMAN’s deep integration with the database engine Enterprise Manager & 3rd Party Tools makes it the best tool for DB backup & recovery • Smart Oracle Database • Sophisticated backup and recovery strategies • Fast • Optimized backup to disk for fastest recovery • No extra redo during backup • Block level incremental backup • Reliable • Block contents validated during backup • Easy Tape Libraries • Simple management with Enterprise Manager • Supports over 20 Media Managers RMAN is used at thousands • Veritas, Legato, Tivoli, HP, Oracle Secure of enterprise sites Backup, etc. • FREE 24 Automated Disk Backup • RMAN fully automates DB backup and recovery to disk • Set and Forget • Nightly incremental backup rolls forward recovery area backup • Changed blocks are tracked in production DB • Full scan is never needed Database Nightly Flash Recovery Archive • Dramatically faster Area Apply Area To Tape Validated • Blocks validated to prevent Incremental corruption of backup copy • Use low cost ATA disk array for recovery area Two Independent Disk Systems 25 Oracle Secure Backup File Systems • Media manager for RMAN backup to Linux, Unix tape of Oracle9i and 10g databases Windows, Filers • Backup file system files • Oracle Home, App Server, etc. Databases • Centralized management of data center backups • Scalability to low 100’s of servers • Best integrated backup of Oracle Databases • End-to-End Encryption of database Supports popular tape backups libraries & drives 26 World’s Fastest Database Backup • Oracle Secure Backup is fast • 10% to 25% faster than competition • Tight integration with Database optimizes performance Leading Vendor • Direct calls into DB engine • Unused Block Elimination skips all unused space • Next release even faster by not Speed of Backup backing up Undo tablespace data 27 Lowest Price Leading Vendor Leading Feature Vendor Oracle Tape Drive $ 3,000 $ 3,000 Client Host $ 1,000 Free Oracle Agent $ 6,000 Free Central Server $ 1,000 Free SAN Backup $ 3,000 Free Encryption $ 2,000 Free Media Server $ 2,000 Free Price NFS Filer-NDMP $ 6,000 Free • Oracle Secure Backup price is just $3000 per tape drive • Backup to virtual tape device (disk) is Free • Low end bundle - one Free direct attached tape drive per database server 28 Sneak Preview of Next Release Backup & Recovery 31 Investigation Lengthens Outages Time to Repair • In an outage, uncertainty and confusion are common • Largest part of downtime is: • Investigating the problem • Choosing the right solution Time to Investigate & Time to plan Total Downtime 32 Data Repair Advisor • Analyzes failures based on symptoms • e.g. “Open failed” because datafiles missing • Intelligently determines repair strategies • Aggregates failures for efficient repair • e.g. for many bad blocks restore entire file • Presents only feasible repair options • Are there backups? • Is there a standby database? • Ranked by repair time and data loss • Can automatically perform repairs 33 Database Backup & Recovery Improvements • Intra-file parallelism for backup and restore of Bigfiles • RMAN duplicate database works across the network • Instantiate standby or test database • No extra temporary copy • Integration with Windows Volume Shadow Service (VSS) Backup Interface • For tape or snapshot backups on Windows 34 Ultra-Safe Mode • Extra protection from changes that corrupt data • Use more hardware cycles to detect and prevent database corruptions • Memory Corruptions • Block Contents Checking • Buffer Integrity Checking – incremental checksum • Disk Corruptions • Lost writes – Data Guard lost write detection • Fractured Writes – ASM sequential mirror writes • ASM Mirroring repairs corrupt blocks from mirror 35 Best Disaster Protection - at Lowest Cost Computer Storage Failure Failures Unplanned Human Error Downtime Data Failures Corruption System Planned Changes Site Failure Downtime Data Changes 36 Best Site Failure Protection - at Lowest Cost Reporting on Production Standby Logical Standby Redo Shipping Database with Real Time Apply Database Data Guard • Best Recovery Point Objective (RPO) - Zero data loss • Best Recovery Time Objective (RTO) – Seconds to failover • Reporting on standby site using Logical Standby • Best Management on Failure - no need to reinstantiate primary after failover • Compared to remote mirroring, Data Guard provides better: • Data protection, network usage, hardware utilization - for free 37 Standby Database Provides Fault Tolerance from Unplanned Failures Computer Storage Failure Failures Unplanned Human Error Downtime Data Failures Corruption • On unplanned failure, failover to standby first • Then take your time to diagnose and repair the failure • Use for cluster failure, corruption, full storage failure, operator error, network failure 38 Automatic Failover to Standby (10.2) Automatic Failover Physical Production or Logical Database Standby DB Synchronous Redo Shipping • Fast-Start Failover implements automatic failover to standby database • Triggered by failure of Site, Hosts, Storage, or Network • Works with and supplements RAC server failover • Failover occurs in seconds • Comparable to cluster failover • Original site automatically rejoins the configuration after recovery 39 Oracle Streams Replication Application Manages Update Conflicts Both Sites Active Redo Shipping • Similar to Data Guard SQL Apply • Shares redo transport, log mining, SQL apply infrastructure • Allows queries and updates on all sites • Application must manage update conflicts • Application also manages role transitions • Failover and resynchronization • Not an out-of-the-box HA solution for packaged applications, but can provide great availability for custom applications 40 Sneak Preview of Next Release Data Guard 41 Physical Standby with Real-Time Query Real-time Queries Concurrent Real-Time Query Continuous Redo Shipment and Apply Primary Physical Standby Database Database • Read-only queries on physical standby concurrent with redo apply • Supports RAC on primary / standby • Queries see transactionally consistent results • Immediate appeal to the many users of physical standby • DR with real time query is unique in the industry – no idle resources • Handles all data types, but not as flexible as logical standby 42 Snapshot Standby – Leverage Standby Database for Testing • Convert Physical Standby to Snapshot Standby and open for writes by testing Physical Standby applications Apply Logs • ALTER DATABASE CONVERT TO SNAPSHOT STANDBY; • Discard testing writes and catch-up to Open Back out primary by applying logs Database Changes • ALTER DATABASE CONVERT TO PHYSICAL STANDBY; • Preserves zero data loss • But no real time query or fast failover Snapshot Standby • Similar to storage snapshots, but: Perform Testing • Provides DR at the same time • Single copy of storage Continuous Redo Shipping • No idle resources 43 Data Guard Improvements • Fast-Start Failover time cut in half • Fast-Start Failover for Maximum Performance (Async) Mode • Automatic failover for long distance standby • May incur small data loss – administrator can limit • Fast Redo Transport • Streaming asynch network protocol • Redo Transport Compression for gap fetching • Fast Incremental Backup on Physical Standby DB • Tracks changed blocks on standby database • New parallel media recovery doubles redo apply performance • SYS user no longer required for redo shipping • Use strong authentication (certificates) 44 Best Data Protection at Lowest Cost Oracle Secure Backup Archive to Tape Data Guard / Streams Site Failure Protection Flash Recovery Area Flashback Corruption Human Error Protection Protection Achieve any Level ASM Mirroring Dramatic Advances of Availability using Storage Failure in Ease of Use Low Cost Hardware Protection 49 Best Online System Changes Computer Failures Unplanned Downtime Data Failures System Online Reconfiguration Planned Changes Online Upgrade Downtime Data Changes 51 Online Reconfiguration – Capacity on Demand • CPU • Add/remove CPUs on SMP online • Cluster Nodes • Add/remove cluster nodes online Database • No data movement needed • Memory • Grow and shrink shared memory and buffer cache online Storage • Auto tuning of memory online • Disk • Add/remove disks online • Automatically rebalance 52 Online Patch Upgrade using RAC Emergency One-off Clients Clients Patch A Patches 1 2 Critical Patch Updates Initial RAC Configuration Clients on A, Patch B Operating System B B Patch Upgrades 4 3 Hardware Upgrades Upgrade Complete Clients on B, Patch A 53 Online Database Upgrades using SQL Apply Upgrade Redo Patch Set Clients Upgrades A B Logs A B Queue Major Version X Version X X X+1 Release 1 2 Upgrades Initial SQL Apply Config Upgrade node B to X+1 Redo Redo Cluster Upgrade A B A B Software & Hardware Upgrades X+1 X+1 X X+1 4 Switchover to B, upgrade A 3 Run in mixed mode to test 54 Sneak Preview of Next Release Database Upgrade 55 Online Patching of One-off Patches • Patch a running Oracle instance with no downtime • Many one-off patches can be online patched • Subset of RAC online upgradeable patches • Great for diagnostic patches • Enable, disable and de-install one-off patches with no downtime • Integrated with OPatch and inventory • Initially available on Linux and Solaris • Long term goal is online patching of Critical Patch Updates (CPUs). 56 Proactive Patching Recommended interim patch becomes available DB/Grid Control Proactively search for patches relevant to customer Patch published on environment eg. by MetaLink with meta feature usage data Apply patch in next Maintenance Window - Identify to DBA Yes Relevant - Download Patch patch Found? Rolling Database Upgrades Using Transient Logical Standby • Logical standby allows rolling Physical upgrades but has data type restrictions • Temporarily convert physical standby to logical to perform a rolling upgrade Logical • Data type restrictions are limited to short upgrade window • No need for separate logical standby Upgrade for upgrade Physical Also possible in 10.2 with more manual steps 58 Best Data Changes Computer Failures Unplanned Downtime Data Failures System Planned Changes Downtime Online Application Data Upgrades Changes 59 Online Table & Index Redefinition • All index changes 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” Copy Transform Source Table Table Result Table GUI Store Update Updates interface Continuous to make it Queries & Tracking Transform Simple Updates Updates 60 Sneak Preview of Next Release Online Application Upgrade 62 Online Application Upgrade • New tools to enable Online Redefinition of Applications • Application developers use these tools in their upgrade scripts • Key Oracle Fusion Applications initiative • Available to all Oracle ISVs and developers 63 In-Place Redefinition Improvements • Fast add column with default value • Does not need to update all rows to default value • Invisible Indexes prevent premature use of newly created indexes • Online index build with NO pause to DML • No recompilation of dependent objects when • Columns added to tables • Procedures added to packages • Easier to execute table DDL operations online • Option to wait for active DML operations instead of aborting 64 Edition-Based Redefinition • Editions provide versioning for database code objects • PL/SQL, triggers, views, synonyms, etc. • Patch and upgrade scripts make changes in a New Edition that is invisible to current production users • New Edition is published for production use after scripts are complete, and new code is tested sql> create Edition “New_Edition” as child of “Old_Edition”; sql> alter session set Current_Edition = “New_Edition”; … Load new versions of non-table objects 65 Versioning a View of a Table • Tables are not versionable in Editions • Create an Editioning View in front of a table to version the table definition • Editioning Views select and name a subset of the columns of a single table • Useful for hiding new or changed columns from production users • Create an editioning view in the “Old Edition” that hides the changes to be made in the “New Edition” • No invalidation, and no transition errors! 66 Features for Handling Data Changes During Upgrades • Set table read-only for all editions • Then copy table and make any changes you want • Set editioning view read-only • No updates allowed in that edition – usually “Old_Edition” • Forward crossedition triggers fire for changes made by sessions in “Old_Edition” • Transform and propagate changes to “New Edition” • Reverse crossedition triggers fire for changes made by sessions in “New Edition” • Transform and propagate changes to “Old Edition” 67 Summary • Best solution for every outage • Best Clustering – Real Application Clusters • Best Human Error Protection - Flashback Error Correction • Best Backup and Recovery – Recovery Manager • Best Disaster Protection – Data Guard • Best Uptime for System Changes – Online Patching and Upgrades • Best Uptime for Data Changes – Redefinition and Editions • Lowest Cost • Low cost grid servers and storage • No idle hardware • Bundled clusterware, volume manager, and DR • Integrated, automated, and simple to use Best Availability AND Lowest Cost 68 Best Practices – 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 • 25 person year effort • otn.oracle.com/deploy/availability Maximum Availability = Unbreakable Architecture + Best Practices 69 71 <Insert Picture Here> Appendix 72 New Oracle Database 10.2 HA Features Clusters Data Guard Generic Diagnostics • Increase Max Instances Automatic Failover Much faster to 1055 on all platforms to Standby db_block_cache_protect • Clusterware Open API Fast Failover to Improved block validation • Clusterware File Standby (seconds) Redundancy for OCR and voting disk Optimized Online Redefinition • Cluster Verification Asynchronous Redo Online Move of a Partition Transmission of a Partitioned Table Data Guard SQL Apply Automatic Deletion Support Clustered Tables • Support for IOTs with of Applied Archive Logs Support Materialized View LOBs and Overflow Logs • Improved Monitoring for Conversion of Physical Standby Support AQ tables SQL Apply • Optimized Creation of DB to Reporting DB Support Abstract Data Logical Standby DB and back types Clone Statistics, Check and Not-null Constraints 73 New Oracle Database 10.2 HA Features Oracle Secure Backup Backup & Recovery Flashback • Tape Management Temporary Datafiles are Re- Flashback Database Support created on RMAN Recovery Through Resetlogs • RMAN Integration Database Transport Across Named Restore • NDMP Same Endian Platforms Points • Fibre Channel Drives Dynamic RMAN Channel Guaranteed Restore Allocation for Backing Up Points • Centralized and Recovering RAC Flashback Across Administration Databases Data Guard • Great Database Backup RMAN HARD compliant Switchovers Performance backups Enhanced RMAN Backup Secure Job Views • Faster Instance Startup Incremental Roll Forward of for Large Memory Database Copy (standby db Configurations or reporting db) 74 New Oracle Database 10.1 HA Features Clusters Data Guard Generic Rolling Upgrades • Portable Clusterware Data Guard Broker Rolling Upgrades Using • Cluster file system for support for RAC Data Guard SQL Apply Linux & Windows Named Data Guard • Automated Patching Configurations Online Redefinition Data Guard SQL Apply Real Time Apply Support of Unique • Support for Longs Flashback Standby Indexes • Support for multi-byte CLOBs and NCLOBs Database One Step Cloning of • Support for Index Flashback Dependent Objects Organized Tables Reinstantiation Columns can be • Functional Indexes Improved Recovery Populated Using • Simplified zero data loss Parallelism Sequences & Sysdate failover Role based Signature Based • Real time apply allows parameter settings Dependency Tracking real time reporting Using Synonyms • Zero downtime Online Segment Shrink instantiation 75 New Oracle Database 10.1 HA Features Automated Disk Based Backup & Recovery Flashback Backup & Recovery Simplified Recovery Flashback Drop • Automated Through Resetlogs Flashback Row History Management of B&R Disk Space Restore Tolerates Missing Flashback Table • Simplified Backup Backups Flashback Transaction Using Image Copy Proxy Backup of Archives History • Change Aware Automated TSPITR Flashback Database Incremental Backups Instantiation Better map of time to • Incrementally Updated Full DB Begin Backup Backups SCN for flashback Automated Backup LogMiner Tuning Channel Failover Automated • Improved Recovery Parallelism Simplified RMAN Specification of Logs cataloging of backup files to Mine • Faster Instance Startup & Cache Warm Automated File Creation Support for Shared during Recovery Server Configurations Drop Database Fine Grained Supplemental Logging Rename Tablespace 76 77 Oracle Streams Replication Application Manages Update Conflicts Both Sites Active Redo Shipping • New Architecture for Replication • Most flexible • Introduced in release 9.2 • Publish/subscribe rules control routing • Best performance • Integrates replication & queuing • Log based – no triggers • Customizable apply transformations & • Much faster than Advanced Rep subsetting • Real-time change capture • Highly available • Zero data loss with remote capture (10.2) • No quiesce for configuration changes 78 Oracle color palette Primary Colors Red Black White Secondary Colors Complementary Colors 79 Photography • Title slides, section dividers and quote call-outs have a place for photography, designated by a grey box with <Insert Picture Here> • A collection of “industry images”, which Oracle has purchased global usage rights for, has been provided for use in these spaces. Additional photography will be made available periodically. • Select images that are appropriate for the specific presentation. “Business” and “Business Lifestyle” images may be used when industry specific images are not appropriate • To insert photography – select <Insert> + <Picture from File> and then scale and position the image exactly over the grey square. See the next slide for instructions on using “the grid” to automatically snap images into place. • The approved library of images can be located at: http://www.oracle.com/wocportal/page/wocprod/corpbrand/graphic_identity/resources/resources_pp.html 80 Using the PowerPoint Grid • When scaling industry images, turn on the “Snap objects to grid” feature with the spacing of 0.083 inches (default) • It is also helpful to choose “Display grid on screen” 81
"Computer Hardware Upgrades"