Improving performance when performing summary over a day and over a time period about aw-server HOT 5 OPEN

I'll start working on it

from aw-server.

@nicolae-stroncea Glad to hear it! Keep us posted.

from aw-server.

This is problematic due to the start and end of a "day" not being well defined because of timezones. One could imagine summarizing over each hour-period, but I'd really like to find another solution as it comes with a lot of complexity.

We've discussed caching like this before, but I think the proper way to improve performance would be to profile the relevant query and find the bottleneck instead of looking into caching straight away.

from aw-server.

What can be done, is that you save the summary of the events at 00:00 daily without the time-zone. So you only write them to the database once. When summarizing over weeks, months, years, all the days in between will still be the same regardless of the time zone as you're essentially summarizing the same chunk of time. However, the difference will be in the first and last day, where the timezone does matter. Therefore you only have to summarize the first and the last day again, by taking the timezone into account, and just add it to the rest of the summarized days, and do analysis from there.

This would not decrease the time to calculate the summary throughout the day, but it would have a significant impact on summarizing over big timelines, which will become even more important as more features are added to analyze productivity.

from aw-server.

@nicolae-stroncea That is true, and we've thought about that before. But as long-term analysis is considered to be a rarely-used function it's not a priority. You described the solution well though, which I don't think anyone has taken the time to write down before, so that is appreciated.

We might get around to it some day, but there are far more pressing issues and the complexity it would contribute to add a cache like that is just not worth the hassle right now.

from aw-server.