Performance Tracking

[Lane]

Was going to post this in Slack, but I think it makes more sense to track it in a thread here. I wanted to have a baseline to compare to for when we do performance tweaks.

Test url: URL: http://eviction-maps.s3-website.us-east-2.amazonaws.com/

Page load on 1040x700 resolution, throttled to “Fast 3G” in OSX Google Chrome:

Cache disabled

• HTTP requests: 49
• Data transferred: 3.5MB
• DOM content loaded: 15.3s
• Load: 16.8s
• Finish: 21.1s

Cache enabled

• HTTP requests: 49
• Data transferred: 3.0KB
• DOM content loaded: 1.98s
• Load: 2.19s
• Finish: 3.75s

[pjsier]

I like this approach! Is the cache disabled in this example? I'm wondering if we should do each test with the cache disabled and then enabled (with a prior page load), since we're leaning on caching a lot here since there's going to be a decent amount of data transfer no matter what

[Lane]

Sounds good, I like the idea of doing cache / no cache. Cache is disabled for the values posted above, will run again and edit with cached values.

[pjsier]

One thing we should also track here is heap size and garbage collection. The biggest performance issue on Firefox is that garbage collection can take a while and slow down everything on the main thread, which is most noticeable when looking at counties at something like zoom 3, and then quickly zooming in. I thought it was a rendering/network issue at first, but the biggest performance issue there is garbage collection on all of the counties that are no longer visible