Last week we talked about the transactions needed for a reporting process. This week’s episode suggested statistics needed about the reports to be produced to properly estimate the reporting process compute capacities. For each report to be produced from the transactional data the key statistics needed include:
- Records Read per Report. How many of the transaction records need to be scanned in total to produce the report. This is a count of the total transactions for the time period covered by the report. For the most efficient reporting processes, this is also the total count of transactions needed to produced all reports from those transactions.
- Records Extracted per Report. This is how many of those records will actually be included in the report. For example, if a report is for a particular business unit, and that business unit is 10% of the total transaction volume for the transaction files to be used for all reports, this is the 10%.
- Bytes Read and Extracted per Report. This is the above statistics converted into record lengths. Reporting processes are driven by both bytes used and record counts, so both statistics are needed for proper estimation.
- Joins Performed per Report. Join processes, to combine data into a usable output, are a major driver of computing resources.
- Frequency of Report. This is how often the report needs to be refreshed or produced.
- Processing Window per Frequency. This is the space in which the report needs to be produced. Often this is a critical element of the daily financial cycle.
- These statistics will be used in the next step of the overall process of estimating reporting processes.
This is Episode 135 of Conversations with Kip, the best financial system vlog there is.