Oracle Data Loader On Demand Guide > Overview of Oracle Data Loader On Demand >

What Factors Affect the Processing Speed?

The queuing mechanism and the throughput rate affect the processing speed and consequently the time it takes to process records.

About the Enhanced Queuing Mechanism

The queuing mechanism allows the server to distribute the resources more efficiently for all users. Thus, a user with a smaller import or data load no longer has to wait for a large import or data load to finish until the small import is processed.

When Oracle Data Loader On Demand submits a request to Oracle CRM On Demand, the file is divided into smaller segments that are processed in parallel by all available tasks on the server. For example, when you submit a file with 4,000 accounts and there are four tasks available on the server, each task processes a data file of 1,000 records. Because of this separation, the server can process the records in any order. For example, records 1001 to 2000 can be processed before records 1 to 1000.

If there are more CSV data segments than available tasks, then the CSV data segments are queued. This is true for all users who submit data using Oracle Data Loader On Demand.

About Throughput Rates

The throughput rate is the number of records processed during a specific period, such as 1 second. You calculate the throughput rate by dividing the total amount of time it takes to process the data by the number of records processed.

What Factors Affect the Throughput Rates?

The following factors directly or indirectly affect the throughput rate of Oracle Data Loader On Demand:

• Type of object (record type). When a record in Oracle CRM On Demand is processed, many records might have to be created or updated. Depending on the complexity of the record type, the number of underlying database records might differ substantially. As a result, the throughput rates differ. The difference is based on the record type that is selected.
• Type of operation (Insert, Update, or Upsert). Throughput rates vary by operation. Update operations are usually faster than Insert operations, because Insert operations might have to write many records and update indexes. Upsert operations are a combination of Insert operations and Update operations and as a result you must include both factors in your estimate.
• Type of fields (picklist, date, description). Some fields require more processing time for validation than others. Picklists, description fields, and dates might process more slowly than Boolean and text fields.
• The number of fields mapped. The more fields you map, the slower the processing of the records. For best results, remove the unnecessary fields.
• Data quality. The more errors in the data file, the more data validation must occur, and more log messages must be written to the log file.
• Log Level. The higher the level of the log file, the more processing time is required, which slows the throughput rate.
• Submission time. Data loads performed during peak hours when other users are also using the server's resources are slower than loads submitted during off-peak hours.
• Associations. When a record is associated with other records, additional processing and validation might be required.
• Background processes. Many background processes are not visible and might affect the throughput rate. These background processes include: log file generation, email notification, and other processes.