Home > Workload Solutions > Oracle > Guides > Reference Architecture Guide—Accelerate Oracle Database using Oracle TimesTen as an Application-Tier Cache > HLR benchmark schema overview
To test the database environment, we created an Oracle Database schema using a purpose-built benchmarking application and workload generator called HLR that simulates a Home Location Register (HLR) type of schema, which is typically seen in the Telco industry.
The HLR schema consists of four tables - SUBSCRIBER, ACCESS_INFO, SPECIAL_FACILITY and CALL_FORWARDING. ACCESS_INFO is a child of SUBSCRIBER, as is SPECIAL_FACILITY. CALL_FORWARDING is a child of SPECIAL_FACILITY. All tables are keyed off the subscriber_id (S_ID column).
The dataset that we created in the RAC database for our testing consisted of 2.5 billion SUBSCRIBER rows, 12.5 billion ACCESS_INFO rows, 12.5 billion SPECIAL_FACILITY rows, and 250 million CALL_FORWARDING rows. The total number of rows in the CALL_FORWARDING table varies slightly as the tests run. This is because the benchmark application performs some INSERT and DELETE database operations against this table, although these operations only make up a small fraction of the overall workload. We created two partitions on all tables and indexes based on the value of S_ID such that each partition contained exactly half the number of rows.
The following table shows the number of SUBSCRIBER rows that were targeted for each use case and the corresponding size in GB when loaded in the respective databases.
Use case |
Target SUBSCRIBER rows |
Target schema and data size in |
|
Oracle database |
TimesTen cache (Physical memory size and type) |
||
Total schema |
2,500 Million |
2,062 GB |
N/A |
Use Case 1 (Small) |
400 Million |
330 GB |
578 GB (768 GB DRAM-only) |
Use Case 2 (Large) |
1,800 Million |
1,485 GB |
2,447 GB (3 TB PMem-MM) |