Db2 high availability in the cloud
Roger Sanders, Principal Offering Manager, Db2, IBM
Published on 10/9/18
Db2 on Cloud is IBMâ€™s on-line transaction processing (OLTP) solution in the public cloud – it allows for rapid deployment of large scale OLTP environments, provides flexible options for both volume and processing speed, and provides a unified architecture that enables hybrid data processing between on-premises and cloud databases. And, Db2 Warehouse on Cloud is IBMâ€™s data warehousing and analytics solution in the cloud. It provides all the capabilities offered by Db2 on Cloud; however, as the name implies, it is optimized for analytical workloads.
With both cloud offerings, an Enterprise High Availability plan is available for environments that require the highest levels of availability. With this plan, a clustered pair of servers are used â€“ one acts as the primary database server and the other serves as a warm standby. In the event of a failure at the primary, workloads will automatically be rerouted to the standby server. Behind the scenes, a special tiebreaker node is used to optimize automated failover decision making and prevent split-brain situations from occurring.
The standby server is kept in sync with the primary server through transaction log record shipping. (More specifically, Db2â€™s High Availability Disaster Recovery (HADR) technology, configured with the NEARSYNC synchronization mode, is used to keep the servers in sync.) Both of these servers are located within the same data center so latency for log shipping is minimal. And, there is a single floating IP address associated with the cluster, which is used to ensure clients/applications are always pointing to the server that is currently acting as the primary.
With this configuration there is currently no service level agreement (SLA) defined. However, the service level object (SLO) is 99.5%, which is the same as that of the non-HA plan IBM offers.
Because log shipping is used to keep the primary and standby in sync, â€śNot loggedâ€ť operations â€“ that is, operations that cannot be effectively replayed on the standby system because there is no transaction logging associated with them â€“ are prevented. Consequently, users are prevented from using CREATE and ALTER TABLE statements with the NOT LOGGED INITIALLY option, as well as the NOT LOGGED option for LOB columns. And, LOAD operations cannot be used to populate the database.
Want to trial Db2 on Cloud? Check out: https://www.ibm.com/analytics/us/en/db2/trials
Join the Hybrid Data Warehouse Community to join in on the conversation: https://community.ibm.com/hybriddatamanagement