Notes

Dear Team,

Please find below the minutes of our recent meeting:

Items completed in this release:

1. Pegging Offloading:
- Completed task: Pegging has been successfully offloaded, resulting in a 40% reduction in memory usage for each scenario. This additional memory can now be utilized to load other scenarios for Dell and PMI.

2. Development of Parallel DTO Read Framework:
- Completed task: The development of the Parallel DTO Read Framework has been finished and delivered.
- Initial implementation: Initially, this feature will be available for two reports:
a. Short Alert Report
b. Push Alert Report

3. Pre-computed Caching:
- Proof of Concept (POC): Pre-computed caching has been implemented as a non-GA feature in this release.
- Caching mechanism: We are utilizing ehcache to cache the results of the reports. It has been proven effective and lightweight.
- Expanded functionality: Previously limited to the web tier, we have extended the caching mechanism to the middle tier. It offers the capability to store data in Heap, off-heap, disk, or any combination of the three.

4. Risks of Caching Pre-computed Results:
- Importance of accuracy: Caching pre-computed results poses the risk of displaying stale data to the user after a planning run. It is crucial to ensure accurate and precise uncaching.

5. Super User Control for Clearing Pre-computed Result Cache:
- Proposed solution: To address any issues that may arise, we will provide super user-level control to clear the pre-computed result cache.

6. Performance Testing:
- Testing scenarios: We performed tests using JMeter with 1, 5, and 10 users simultaneously retrieving data.
- Results awaited: The performance and stability of the system under different loads are currently being analyzed.

7. Impact on CBP Runtime:
- Increased runtime: Due to pre-computation, the CBP runtime has increased by 42 seconds. However, this increase is within acceptable limits.

8. POC Implementation and Statistics:
- Reports tested: We implemented and collected statistics for the following reports:
a. Short Alert Report
b. Delivery Plan Report

Next Plan:

1. Benchmarking ehcache Performance:
- Objective: We will conduct benchmarking to evaluate the performance of ehcache with 10 million records. Different configurations of Heap, off-heap, and disk storage will be tested.

2. Timeline for Critical Reports:
- Expected timeline: All critical reports, totaling around 15, will be incorporated with the pre-computed result caching feature within one release.

3. Limitation to Planning Reports:
- Scope clarification: The pre-computation and caching features will only be implemented for planning reports, excluding master reports.

4. Heap Memory Consideration:
- Heap memory impact: Implementing caching for all 15 reports and storing everything in Heap will likely increase the memory usage. This aspect requires careful evaluation and planning.

5. Performance Improvement for Cockpit:
- High priority: Pre-computation and caching of reports are regarded as high priority items for version 23.4, as they will significantly enhance the performance of the cockpit.

6. Parallel DTO for PS:
- Handover to PS: The Parallel DTO functionality will be provided to the PS team, enabling them to explore its capabilities.

7. CBP Optimization for JLR:
- Current CBP runtime: The current CBP process, involving 2000 critical paths and 52 weeks of planning, takes approximately 40 minutes for JLR.
- Root cause analysis: A line item will be created to investigate the underlying reasons for the prolonged CBP runtime.

8. Disk Tier Usage for Cache Storage:
- Possibility of disk tier: Considering that SSDs are used in production boxes, we can explore utilizing the disk tier for storing cache results.

Open Questions:

1. Asynchronous Computation and Synchronous Cache Clearing:
- Exploration needed: We can explore the possibility of parallelizing the computation and caching processes for the 15 reports, making it asynchronous. However, clearing the cache after a planning run might need to remain synchronous. Further discussions are required.

2. ehcache Configuration at Report Level:
- Feasibility inquiry: Currently, ehcache configuration is set at the scenario level. It needs to be determined if it is feasible to configure it at the report level. Consideration should be given to code maintainability and development best practices.

3. Extensive Testing with Different Records:
- Expanded testing: Can we conduct testing with more records, such as PMI or Dell, to verify the effectiveness of ehcache with these configurations? Further analysis is required.

Please review the minutes, and if you have any additional points to add or any modifications, kindly let me know.