When APO first came out, we enabled it on our site expecting some decent performance increases. Our Front page is entirely static, and where most of our traffic goes to. We only have a single physical server hosting the front page, located in New York City. However, it seems it massively increased “Download” and Time to first byte.
These are all charts of our front page statistics exclusively. First ones are from Datadog Synthetics.
We use every Datadog monitoring location, and overall turning it off a few hours ago instantly put both download and TTFB down to around 80-100ms from ~500ish.
It doesn’t seem to matter what location it was either. Interesting, because I thought APO would help with far-away locations from the server as it would remove round trips to the origin itself.
This is of North California, much closer to the physical server, but still around the same results.
These are graphs for the last 8 months. First we tested APO around November, and then in December started using it fully. You can see the download time going way up.
Here’s a screenshot of Cloudflare’s own health checks, can see the decline the second APO was disabled.
I thought this was really interesting. I believe APO runs off Cloudflare Workers and KV, instead of just normal Cloudflare Cache, so I would expect maybe a bit worse performance, but not this bad. Hopefully, these graphs/info helps someone.
Anyone else have experiences like this with APO? It seems that it would be more efficient to just clear cache manually when updating site.
Thanks for your time
Edit: Opps, I marked as solve before replying and can’t reply anymore.
Thanks to the Cloudflare team for looking into this and providing us more insight into this. APO is really cool, and I totally understand it isn’t a magical silver bullet. Thanks for your time!