| Bookshelf Home | Contents | Index | Search | PDF | |
Siebel Analytics Performance Tuning Guide > The ETL Process > Rearrange Batches for Balance >
Example
Suppose the company Data Warehouse Inc. uses the FactStaging2 and FactStaging3 batches to load their data warehouse. These batches are shipped with several sessions within each batch as shown in Figure 6.
Their Activity Fact, Order Item Fact, Service Request, and Revenue Fact tables have several million rows, and the rest of the fact tables have a few hundred rows.
The loads for the multimillion row fact tables each required nearly four hours (the Service Request required four and a half hours), whereas the other loads required less than thirty minutes per load.
When you run the batches as shown above, the FactStating2 batch requires four hours, the FactStagins3 batch requires about four and a half hours: eight and a half hours for two batches. The Batch FactStaging2 is running, Activity Fact and Order Item is running for long time while the other sessions took less than 30 minutes which means the load on the Informatica server is very minimal.
In a similar way, if you see the FactStaging3 Batch is running, the Service Request and Revenue Fact ran for long time, and the other session took less than 30 minutes.
You can rearrange the sessions in these two batches so that the Informatica Server can be used more efficiently and therefore save time. You can place the sessions that took a long time in one batch and move all the other sessions into another batch.
For example, you could move the Service Request and Revenue Fact into the Fact Staging2 and the rest of the sessions into FactStaging3 as shown in Figure 7.
After rearranging the sessions, the FactStaging2 batch required five hours and the FactStaging3 required about a half hour.
| Bookshelf Home | Contents | Index | Search | PDF | |
Siebel Analytics Performance Tuning Guide Published: 18 April 2003 |
