High Performance Computing

 View Only

Sharing the same database with multiple IBM Spectrum Symphony clusters

By Jimmy Liu posted Tue October 31, 2023 04:14 PM

  

For IBM Spectrum Symphony clusters, the PERF reporting database schema cannot be shared across multiple clusters. For optimal security, create multiple database instance for each IBM Spectrum Symphony cluster to isolate data securely for each IBM Spectrum Symphony cluster. This way, there is no sharing of data between each IBM Spectrum Symphony cluster. However, if you can’t create multiple database instances for each cluster, you can share the same database server instance, which is still a very secure option. This blog post provides suggestions and tips for configuring a shared DB2, Oracle, or SQL Server database instance between multiple IBM Spectrum Symphony clusters.

Here are examples for configuring shared database instances; the guidance is different, depending on the type of databases: IBM DB2, Oracle, and Microsoft SQL Server.

For DB2, create a different schema for each cluster, and then create IBM Spectrum Symphony tables, views, and so on, under each schema. This way, one schema owns one IBM Spectrum Symphony cluster’s data. If there is data sharing between different IBM Spectrum Symphony clusters, then create different DB2 users and grant full privileges to each schema. If you want to share data between different clusters, depending on security, if two clusters can be fully shared, then two clusters schema can share same DB2 user by granting full privileges to one user to own both the IBM Spectrum Symphony cluster data schemas. If you want a more secure environment with separate cluster data, but still share some level of data, you can create a new DB2 user, grant certain access privileges to different schemas, and then generate a joint report from this new user.

For Oracle, create different users for each cluster, and then create IBM Spectrum Symphony schema under each user. This way, one Oracle user owns one IBM Spectrum Symphony cluster schema. If want to share data between different clusters (for example, joining multiple clusters’ data and generating an aggregated report), this can be done by creating a new, separate user, and then granting this user privileges to access the required joint cluster’s user schema. An aggregated report can be generated from this new separate user, who has privileges to access multiple clusters’ data.

For Microsoft SQL Server, create different users so that each user owns a different schema for each cluster, and then create IBM Spectrum Symphony tables, views, and so on, under each schema. This way, one schema loads one IBM Spectrum Symphony cluster’s data. If you want to share data between different clusters, depending on security, if two clusters can be fully shared, then two clusters schema owners can be set to the same SQL server user, so that this user owns database schemas for data for two IBM Spectrum Symphony clusters. For more security, and separate cluster data, but still sharing some level of data, create a new SQL Server user, grant certain access privileges to different schemas, and generate joint reports from this new user.

Regardless of which type of database you use, for best performance and security, a DBA needs to design and configure the database architecture for multiple clusters, and estimate the data volume if multiple clusters share the same database instance.

0 comments
29 views

Permalink