Thoughts on: Big Data, Hadoop, Databases, Platform, Performance & Scalability
Real-World Performance Videos on YouTube – OLTP
In addition, here are some OLTP demos that demonstrate how much performance and throughput can be wasted by poor design and suboptimal database programming.
Nice examples. Cool to see the examples and impacts side by side. I of course enjoy seeing the OEM graphics helping to convey the impact as well.
- Kyle
[...] Greg Rahn directs towards some OLTP demos that demonstrate how much performance and throughput can be wasted by poor design and suboptimal database programming. [...]
Nice to see the concept explained in action through video, kind of drives the point home in 10 minutes. It is sad that even after 10+ years of advocacy by experts like you there are still developers out there who don’t get it. Avoiding parsing should be taught as part of database programming 101.
A question about information from second video. We know that best target for hard parse is zero. Without trial and error how can we find out optimal number of connections in the connection pool? Or the only way is to run the actual workload and try various values while monitoring with a tool like OEM (or running queries against ASH)?
Most of this is lab work – it’s different for every application. There should be some stats on the JDBC connection pool that you can reference as a starting point.
Nice examples. Cool to see the examples and impacts side by side. I of course enjoy seeing the OEM graphics helping to convey the impact as well.
- Kyle
Great presentation…
Thanks for sharing this.
[...] Greg Rahn directs towards some OLTP demos that demonstrate how much performance and throughput can be wasted by poor design and suboptimal database programming. [...]
Very Nice videos. what is that tool that is being used in the video?
The tool is something we wrote for Exadata Database Machines that is a “pretty printer” on top of collectl.
Nice to see the concept explained in action through video, kind of drives the point home in 10 minutes. It is sad that even after 10+ years of advocacy by experts like you there are still developers out there who don’t get it. Avoiding parsing should be taught as part of database programming 101.
A question about information from second video. We know that best target for hard parse is zero. Without trial and error how can we find out optimal number of connections in the connection pool? Or the only way is to run the actual workload and try various values while monitoring with a tool like OEM (or running queries against ASH)?
Thanks
Most of this is lab work – it’s different for every application. There should be some stats on the JDBC connection pool that you can reference as a starting point.
[...] DWH video [...]