July 16th nightly numbers are captured in attached document and shows 190 ms regression for DS in both cached and no cached flow when compared with AS on MV reference hardware. Capturing things to try and optimize DS first load 1. reduce feed data via domain_affinities and by how much. How does this impact personalization vs performance. - Ruled out for now 2. remove optimistic loading and test, how much do we gain from this. - Punam/Jeane - ( Done Numbers captured in doc) 3. fixed images weight, how much of the noise we're seeing is due to images being loaded in. - Punam/Jeane 4. confirm feeds are still gzipped (and why cache is also regressing) - Gavin 5. JS profiling - Scott 6. test old hardcoded layout with topics (68) vs current (69). 7. Figure out how much of this is regression and how much is noise. 8.Beta matrix with first frame to validate regression in beta- Jeane 9.Image compression optimization- Gavin 10. Run number With pocket adzerk server endpoint reverted to 68 to rule out regression due to new spoc endpoint - https://github.com/mozilla/activity-stream/commit/7a795a41c561156ba754afe28d0e68e4a01024f2#diff-8b1b03c4d5dec5c8e4f7de50e05c0c4f
Bug 1564223 Comment 3 Edit History
Note: The actual edited comment in the bug view page will always show the original commenter’s name and original timestamp.
July 16th nightly numbers are captured in attached document and shows 190 ms regression for DS in both cached and no cached flow when compared with AS on MV reference hardware. Capturing things to try and optimize DS first load 1. reduce feed data via domain_affinities and by how much. How does this impact personalization vs performance. - Ruled out for now 2. remove optimistic loading and test, how much do we gain from this. - Punam/Jeane - ( Done Numbers captured in doc) 3. fixed images weight, how much of the noise we're seeing is due to images being loaded in. - Punam/Jeane 4. confirm feeds are still gzipped (and why cache is also regressing) - Gavin 5. JS profiling - Scott 6. test old hardcoded layout with topics (68) vs current (69). 7. Figure out how much of this is regression and how much is noise. 8. Beta matrix with first frame to validate regression in beta- Jeane 9. Image compression optimization- Gavin 10. Run number With pocket adzerk server endpoint reverted to 68 to rule out regression due to new spoc endpoint - https://github.com/mozilla/activity-stream/commit/7a795a41c561156ba754afe28d0e68e4a01024f2#diff-8b1b03c4d5dec5c8e4f7de50e05c0c4f