230 likes | 388 Views
E148 Achieving 24x7 Availability. Rick Linden Product Manager ESD rick.linden@sybase.com. Strategies for Database Availability. Switching and Warm Standby Replication. Cold Standby. Disaster Recovery. Catastrophic. High Availability Clusters. High Availability. Unplanned.
E N D
E148Achieving 24x7 Availability • Rick Linden • Product Manager • ESD • rick.linden@sybase.com
Strategies for Database Availability Switchingand WarmStandbyReplication ColdStandby DisasterRecovery Catastrophic HighAvailabilityClusters HighAvailability Unplanned Severity of Database Downtime OfflineMaintenance Planned OnlineMaintenance No Downtime ContinuousAvailability Latency of Database Recovery
Dynamic Reconfiguration Dynamic reconfiguration and self-tuning capability will reduce server downtime and will save server maintenance costs. Key Benefits: • Need to reboot server is reduced, increasing server up-time • Automatically adjust configuration parameters to deal with server load changes reducing DBA efforts • Dynamic Tuning
Dynamic Reconfig params: Total memory Number of user connections Number of locks Number of open objects Procedure Cache (absolute) Disk I/O structures Online index Monitoring agent will assess the health of ASE using Systems Mgm API and automatically apply configuration changes Dynamic Reconfiguration Solution in ASE 12.5 Self-Tuning assisted by Dynamic Re-configuration Load ASE SM DynamicRe-configuration Monitoring Agent
Dynamic Reconfiguration Compressed Back-ups Decision Support Quiesce Database Nears instantaneous mirror splits with partner technology Online maintenance
Quiesce Database Secondary Primary 2:00 AMquiesce database hold;<copy database usingexternal command>;quiesce database release 7:00 AMdump tran with standby_access 9:00 AMdump tran with standby_access 10:00 AMdump tran with standby_access Repeat each houruntil activity tapers off; then lengthen intervals accordingly 2:10 AMdataserver-q .. 7:05 AMload tran;online database for standby_access 9:07 AMload tran;online database for standby_access 10:10 AMload tran;online database for standby_access Warm Stand by Model using Quiesce Database
Companion Overview • Two ASEs running on separate nodes • No server startup time • Works on the top of hardware subsystem for sharing disk and process information • HA subsystem detects failure and transfers control of disks and connections to the surviving ASE • Works with Sun, HP (Compaq), IBM, MSCS • Client failover and failback capability • Supports Sun Cluster and Veritas Cluster Server
High Availability Clusters Switchingand WarmStandbyReplication ColdStandby DisasterRecovery Catastrophic HighAvailabilityClusters HighAvailability Unplanned Severity of Database Downtime OfflineMaintenance Planned OnlineMaintenance No Downtime ContinuousAvailability Latency of Database Recovery
Replicate Users/Logins Cluster Availability HA System Establish S1 S2 Companion Node 2 Node 1 Shared Disk Storage Disk Disk
Fail Over Cluster Availability HA System S2 Node 2 Shared Disk Storage Disk Disk
Fail Back PrepareFailback Cluster Availability HA System S2 Node 2 Shared Disk Storage Disk Disk
Fail Back Replicate Users/Logins Cluster Availability HA System Establish S1 S2 Companion Node 2 Node 1 Shared Disk Storage Disk Disk
Summary • Dynamic Re-Configuration • Sybase HA addressed Planned and unplanned downtime • Online maintenance • Quiesce Database • Companion server • No client disconnects during HA failover • No data loss • Transparent client fail over
Strategies for Database Availability Switchingand WarmStandbyReplication ColdStandby DisasterRecovery Catastrophic HighAvailabilityClusters HighAvailability Strategies for Database Availability Unplanned Severity of Database Downtime OfflineMaintenance Planned OnlineMaintenance No Downtime ContinuousAvailability Latency of Database Recovery
Disasters beyond systems • Physical site loss because of natural disasters, power outage for area, and other catastrophic losses • Dire consequences occur due to these business interruptions Costs • Just think of the cost of days of downtime because of a disaster - $2,500-$10,000 a minute – Millions a day • In addition, there is the potential lost of customers who can’t access your systems
Sybase Solution Degree of Interruption/Latency of Data ASE 12.5 Planned ASE 12.5 HA Replication & OpenSwitch HA Clusters Unplanned Challenges ASE Backup & File Transfer Cold Mirroring Disaster Update based on one from Phils paper on page 12
Standby DB OpenSwitch Overall architecture App Primary DB Warm Standby Replication Server
Replication Server/Open SwitchValue Proposition Provides the only solution that address all three areas of HA Key Benefits • Works across the LAN and WAN - no geographic restrictions • High performance replication • Database Integrity is guaranteed • Redundant hardware can be leveraged for DSS or load balancing • Low setup and maintenance cost • Automates the failover switch
Warm Standby Features • Provides highly available data with minimal latency • Includes schema replication • No application impact • Easy configuration and fast to market: • No replication definitions or subscriptions required • Low setup and maintenance costs • Simple switch to standby on DB failure • Useful for: (this may belong on another slide) • load balancing for decision support systems • fault tolerance • disaster recovery
Open Switch Features • Provides functionality to manually or automatically manage client failover • Includes error detection and recovery features • Can manage any number of remote servers • Integrates into any HA environment • No change to existing application
Active DB Standby DB OpenSwitch RS New Active Warm Standby with Open Switch Queue
How solution fits the problem The only solution that addresses all three areas of HA • A geographically separated copy is critical for Disasters (or site failures) • RS Warm Standby & Open Switch work across WAN and Warm Standby maintains a transactional consistent standby copy • Quick Fail-over and Fail-back procedures are critical for Unplanned downtime • Open Switch provides seamless client fail-over and fail-back • Warm Standby allows you to switch the direction of replication on failure or fail-back which is critical for resynchronization • Warm Standby can minimize planned downtime and/or provide data availability during maintenance downtime
Database Availability Summary Switchingand WarmStandbyReplication ColdStandby DisasterRecovery Catastrophic HighAvailabilityClusters HighAvailability Unplanned Severity of Database Downtime OfflineMaintenance Planned OnlineMaintenance No Downtime ContinuousAvailability Latency of Database Recovery