Improve performance

[elklaassen]

At the OK Lab in January, we became several suggestions:

• Pre calculate data (implemented to some extend, further optimization possible)
• Apply ETL (Extract, Transform, Load), add lates hour/date data to database
• Measure performance using browser tools (F12)
• Check server/client side load
• Use PyCharm profile methods (checked, this is a paid service only)
• Use Airflow
• Use KNIME (checked, seems to be a dahsboard tool whereas we opted for python/dash/plotly)
• Use Pandas and pre-compute (already implemented)
• Use Timescale
• Check @functools, "poespas", lazy loading
• Use Tabs (this would come in V2.0 to structure functionality anyhow)

Further suggestions welcome.

[elklaassen]

I reviewed performance and made the following notes for further investigation:

• Initial load takes ca. 4.9 sec. of which 4.6 sec. "Waiting for server response". This involves main.css, custom.css and dash extensions. I wonder why we need custom.css (greyscale)? DashExtensions_default.js appears to refer to telraam eco counter code?
• Next: dash_layout needs ca. 4.8 sec. of which 4.5 sec. "Waiting for server response" . Same for dash_dependencies.
• Next: ca. 4.7 sec for Plotly and components.
• Parallel to previous: dash_update_components (2x) need 12 sec. incl. 7 sec. "stalled"
• Update street: pie chart 6.6 sec. of which 6.2 sec. "Waiting for server response", similar for street map update.
• Update Graph radio button, similar as above...