Any business interruption is a potential loss of revenue. Achieving business continuity involves a tradeoff between the cost of an outage or data loss with the investment required for achieving the recovery point objective (RPO) and recovery time objective (RTO). Continuous system availability requires scalability, as well as failover capability for maintenance, outages, and disasters. It also requires a shift from standby to active-active systems. Active-active sites are geographically distant transaction processing centers, each with the infrastructure to run business operations and with data synchronized by using database replication, such as the Q Replication technology that is part of IBM® InfoSphere® Data Replication software. This IBM Redbooks® publication describes preferred practices and introduces an architecture for continuous availability and disaster recovery that is used by a very large business institution that runs its core business on IBM DB2® for z/OS® databases. This paper explains the technologies and procedures that are required for the implementation of an active-active sites architecture. It also explains an innovative procedure for major IT upgrades that uses Q Replication for DB2 on z/OS, Multi-site Workload Lifeline, and Peer-to-Peer Remote Copy/Extended Distance (PPRC-XD). This paper is of value to decision makers, such as executive and IT architects, and to database administrators who are responsible for design and implementation of the solution.
Understanding the impact of workload and database characteristics on the performance of both DB2®, MQ, and the replication process is useful for achieving optimal performance.Although existing applications cannot generally be modified, this knowledge is essential for properly tuning MQ and Q Replication and for developing best practices for future application development and database design. It also helps with estimating performance objectives that take these considerations into account. Performance metrics, such as rows per second, are useful but imperfect. How large is a row? It is intuitively, and correctly, obvious that replicating small DB2 rows, such as 100 bytes long, takes fewer resources and is more efficient than replicating DB2 rows that are tens of thousand bytes long. Larger rows create more work in each component of the replication process. The more bytes there are to read from the DB2 log, makes more bytes to transmit over the network and to update in DB2 at the target. Now, how complex is the table definition? Does DB2 have to maintain several unique indexes each time a row is changed in that table? The same argument applies to transaction size: committing each row change to DB2 as opposed to committing, say, every 500 rows also means more work in each component along the replication process. This RedpaperTM reports results and lessons learned from performance testing at the IBM® laboratories, and it provides configuration and tuning recommendations for DB2, Q Replication, and MQ. The application workload and database characteristics studied include transaction size, table schema complexity, and DB2 data type.
Understanding the impact of workload and database characteristics on the performance of both DB2®, MQ, and the replication process is useful for achieving optimal performance.Although existing applications cannot generally be modified, this knowledge is essential for properly tuning MQ and Q Replication and for developing best practices for future application development and database design. It also helps with estimating performance objectives that take these considerations into account. Performance metrics, such as rows per second, are useful but imperfect. How large is a row? It is intuitively, and correctly, obvious that replicating small DB2 rows, such as 100 bytes long, takes fewer resources and is more efficient than replicating DB2 rows that are tens of thousand bytes long. Larger rows create more work in each component of the replication process. The more bytes there are to read from the DB2 log, makes more bytes to transmit over the network and to update in DB2 at the target. Now, how complex is the table definition? Does DB2 have to maintain several unique indexes each time a row is changed in that table? The same argument applies to transaction size: committing each row change to DB2 as opposed to committing, say, every 500 rows also means more work in each component along the replication process. This RedpaperTM reports results and lessons learned from performance testing at the IBM® laboratories, and it provides configuration and tuning recommendations for DB2, Q Replication, and MQ. The application workload and database characteristics studied include transaction size, table schema complexity, and DB2 data type.
Any business interruption is a potential loss of revenue. Achieving business continuity involves a tradeoff between the cost of an outage or data loss with the investment required for achieving the recovery point objective (RPO) and recovery time objective (RTO). Continuous system availability requires scalability, as well as failover capability for maintenance, outages, and disasters. It also requires a shift from standby to active-active systems. Active-active sites are geographically distant transaction processing centers, each with the infrastructure to run business operations and with data synchronized by using database replication, such as the Q Replication technology that is part of IBM® InfoSphere® Data Replication software. This IBM Redbooks® publication describes preferred practices and introduces an architecture for continuous availability and disaster recovery that is used by a very large business institution that runs its core business on IBM DB2® for z/OS® databases. This paper explains the technologies and procedures that are required for the implementation of an active-active sites architecture. It also explains an innovative procedure for major IT upgrades that uses Q Replication for DB2 on z/OS, Multi-site Workload Lifeline, and Peer-to-Peer Remote Copy/Extended Distance (PPRC-XD). This paper is of value to decision makers, such as executive and IT architects, and to database administrators who are responsible for design and implementation of the solution.
The Communist Party appeared a hundred years ago on the French political and social scene. According to opinions and moments, it has been the party of Moscow, of those shot, of the working class, of the union of the left, the party of the foreigner or that of the nation. It has been underground, in government, in town halls, in factories or in the streets. Some considered it too revolutionary, others not enough. More than others, it aroused passions, positive or negative. It attracted many and repelled just as many. After the fall of the USSR, it decided to remain a communist party, while many others gave it up. But it no longer has the place it once had, in reality as in the imagination. This book does not intend to judge, but to provide keys to understanding. It is based on a considerable number of archives that are now available and is an ordered and distanced look at an object that is not lacking in complexity and no doubt even in mystery. This book has been translated from French to English thanks to a financial help from the Gabriel Péri Foundation and the LIR3S UMR Cnrs 7366 of Dijon.
Thank you for visiting our website. Would you like to provide feedback on how we could improve your experience?
This site does not use any third party cookies with one exception — it uses cookies from Google to deliver its services and to analyze traffic.Learn More.