- Hyper avg extract execution is 30% faster than TDE
- Hyper avg extract size is 10% larger than TDE
- Hyper view avg render time is 15% faster than TDE view render time
Interesting.... Ritesh's is consistently 'smaller' than mine for all the areas. Here is my findings again.... Definitely I do not see 3X improvements. Love to see more observations from others....
- Hyper avg extract execution is 50% faster than TDE
- Hyper avg extract size is 20% larger than TDE
- Hyper view avg render time is 20% faster than TDE view render time
I am running 10.5.2 and 10.5.3 instances
- Hyper avg execution is not faster it is actually slower by about 20%
- Hyper avg extract size is 20%larger on extracts w.o comments, with comments it close to 40% larger.
- Hyper on EC2 m5.4xlarge- extracts get corrupted with each tabcmd refresh or publish as well as with each scheduled refresh . Case opened with tech support, no response
- Hyper on EC2 r4.4xlarge - extracts get corrupted via tabcmd refresh or publish if record count in extract is greater than 20 million Case opened with tech support, no response
Are you still seeing slower execution of the hyper extract comparted to the TDE? I don't have exact numbers, but am seeing significantly slower performance on 10.5.2. Wondering if you were able to resolve and get the faster expected performance.
I am working on the difference between different extract formats. A significative example is an extract of 410,413 row:
- .CSV size 48.546 Mb
- .Hyper size 1.856 Mb
- .tde size 3.224 Mb
The render times are the same between .hyper and .tde, while .csv is slower. I don't want to write the percentage of render time because, for a small amount of rows, the percentage could be unreliable.