Tuesday, February 02, 2016

investigating Delphix job timeout issue with inner tools




Recently we have Delphix jobs can't  complete within scheduled windows. even I separate them in more windows. 

The above three pages are handy for investigation. They tell me which databases are using the most bandwidth. From them drill down further, I identified problematic application/jobs, which should not be there. I even found  a problem in production, which does full table scan to a moderate size table. This is make the saying true that fast HW may help to hide the performance problem. 

As we use NFS to present storage to database, the network bandwidth becomes expensive for our 1GbE network interface , running on VMWare. 

From this case, we learn that for best practice, the recommend network is 10GbE


Best practices to improve network throughput include:
Use 10 Gigabit Ethernet (10GbE)
Use a dedicated storage network

And it is worth to read the Delphix performance tuning document again.