Situation: Data loading process failure! – Oct 2013
The bulk data load job got stuck due to a different process which utilizes the same table. The bulk data load was a high priority and the Client was very upset since the business needed the data live due to a product launch.The data in queue was around 10,00,000 records. The Client gave a strict extended deadline of 2 hours!
Solution: Stopped the low priority jobs running on database triggered by team’s user to avoid any locks on the tables concerned.Checked the number of available sessions on server using PLSQL developer.Took 50% of all the sessions available on the servers. Divided the data into small batches and triggered the data loading process in parallel across all the servers. The data was loaded successfully in 1.5 hours!
Situation: Data loading job affected due to server crashes! – Sept 2013
The sim registration axion job which loads data got affected because the server on which the job was running used to crash frequently
Solution: Modified the Data loading job. Created a Flip – Flop process where two same jobs will be triggered with a time difference of half hour with a batch check in – check out process which ensured that there is no duplicate data processing. Also the process ensured that the batch is processed before checking the data out and checking in the next batch. Previously the data loading process used to take 3 hours for a batch of 5000 records. Due to the parallel process implementation the processing capability increased to around 9000 per 3 hours which also drastically reduced the ticket count.
Situation: Unregistered customers due to invalid names and missing attributes! – June – 2013
There was no data validation on First Name and Last Name fields and also business introduced some more attributes, the records of which were missing for previously registered customers. The count of such customers was around 600,000.
Solution: Collected the contact details of customers whose attributes are missing. Using a SQL procedure, collected the details of customers who have invalid first name and last names. Using a email and phone campaign contacted the customers to get them registered properly with all the details.
Situation: Data out of sync between CRM (My Team) and Postpaid team! – Oct 2012
The weekly sync process between CRM and Postpaid team failed which resulted in a huge data backlog on CRM. The initial process was manual procedure whenever there are postpaid registrations.
Solution: Created a Unix shell script to automate the entire data loading process. The process every week compares the data between CRM and Postpaid and if there is any difference, it will automatically trigger the data load job to load the missing data.
Rohan Ashok Patil 