Application Considerations

Review the following:

• General Recommendations
  
• Oracle Healthcare Foundation
  
• Cohort Data Model
  
• Omics Loaders
  

General Recommendations

• Oracle Table or Schema Statistics:

  Configure the scheduled jobs to collect the Oracle statistics using the following method. This should be configured after the initial load is complete.

  eBEGINDBMS_SCHEDULER.create_job ( job_name => 'OWNER_STATS', job_type => 'PLSQL_BLOCK', job_action => q'[BEGIN dbms_stats.gather_schema_stats (ownname=> 'OWNER', method_opt=> 'for all columns size auto', estimate_percent=> dbms_ stats.auto_sample_ size, cascade=>true, block_sample => true, options => 'GATHER STALE'); END;]', start_date => sysdate, repeat_interval => 'FREQ=DAILY;BYHOUR=0', end_date => NULL, enabled => TRUE, comments => 'Table stats gathering job for OWNER.');END;/

• Data files or flat files (for example, Oracle Database result and reference data files) should be staged on Database File System (DBFS) as the Oracle Database provides much better security, availability, robustness, transactions, and scalability than the traditional file systems. For setting up DBFS, see article ID 1054431.1 on My Oracle Support (https://support.oracle.com).

Oracle Healthcare Foundation

• Oracle Table or Schema Statistics:

  HDI source tables should have proper table statistics before the initial load. See Oracle Healthcare Foundation Administrator's Guide for details on gathering statistics and for ETL-specific recommendations regarding parallel options.

• Oracle Healthcare Foundation Data loads:
  • When you load HDI data, make sure that you load data using the direct path insert to obtain a high compression ratio. For more information about improving INSERT performance with Direct-Path INSERT, see the Oracle Argus Insight Database Administrator's Guide, and implement accordingly for your ETL loads to HDI.
  • If direct path insert is not possible due to code or ETL limitations, rebuild the tables or partitions using the alter table .. move .. SQL command (indexes on these tables should be rebuilt) to achieve better compression. However, this leads to regular maintenance of these tables to achieve maximum compression. Oracle recommends that you create scripts or ETL to support direct path insert rather than resorting to regular maintenance. For more details, see Oracle Healthcare Foundation Programmer's Guide.

Cohort Data Model

See Oracle Health Sciences Translational Research Center Installation Guide for Exadata-specific configurable partitioning and parallel options.

Omics Loaders

• Run the result files in parallel to achieve maximum throughput.

  Note:

  You can execute maximum 4gVCF loaders on a single Exa node.
• See Data Model Installation for Exadata-specific configurable partitioning and parallel options.