Oracle Real Application Clusters

During this process the organization needs to determine the business requirements, such as availability, scalability, and all other requirements defined in Chapter 1. If the organization determines that all these business requirements are critical to its everyday functioning as an enterprise for continued business and customer support, it would be ideal to investigate whether the current application could be migrated to a clustered solution.

A basic question is, has the application been designed for a clustered database solution? This was truer in the case of migrating from a single stand-alone configuration to the previous version of the clustered database platform OPS. With RAC, this requirement is of less importance. The main reason, as we have seen from the previous chapters, is because all the communications occur across the cluster interconnect.

Under OPS, all request sharing happened via processes pinging the information from disk, and in RAC this happens via the cluster interconnect. Due to the heavy disk activity, it is a requirement when implementing a clustered database solution with OPS that the application has specific design considerations that would lessen such high activity.

Scalability of the applications, as mentioned in Chapter 1 (Section 1.2.7), is an important database requirement. One of the database features that support this requirement is the data-partitioning feature. While partitioning is not a requirement for RAC, it would help in data distribution and therefore provides scalability. If the current database has already taken advantage of the data-partitioning feature or if the database implementation is changed to use the data-partitioning feature, it would be beneficial to the overall performance of the application.

Certain applications provide homegrown consistency locking mechan isms. RAC, with the clustered architecture, provides its own mechanism of lock management when users from more than one instance request the same information. It should be analyzed thoroughly to ensure that these home-grown locking solutions would function in a clustered database environment.

During this analysis process, one of the questions that should get answered is, why should you migrate to a clustered solution? If all the business requirements stipulated can be met by the existing configuration, or if a small change to the current configuration can help meet the requirement, would that be sufficient? This is a very important question, because when the cost-benefit analysis is determined before implementa tion, this will provide a most convincing answer. On the other hand, if this is not answered when the system is implemented and alive, when the real benefits expected are not released, it could be too late and a huge investment could be wasted.

The next major step before the actual migration to this new platform is to determine the actual cost-benefit analysis.