Let's have a quick (long?) chat about web performance.
I have the unique situation of seeing data from billions of performance profiles as we help customers with @raygunio's Real User Monitoring look to improve user experience. #webdev #PerformanceMatters #webperf
-
Show this thread
-
1. Many folks still think a synthetic test is a Good Thing. That is, pinging a website from a fixed set of servers. It's better than naught, but it's a pretty poor proxy for real customer experience. You're measuring data center to data center performance (even if nerfing perf)
1 reply 0 retweets 1 likeShow this thread -
2. Even with nerfing perf, you're talking high quality interconnects between DCs. This compares poorly, to say, what an actual user experiences on a 3G phone while roaming around somewhere. You're not getting realistic numbers.pic.twitter.com/4SYUFZlvoC
1 reply 0 retweets 2 likesShow this thread -
3. Your only real way to understand performance is to track what the end users load time actually is (hence the 'Real User' in 'Real User Monitoring'). I'm surprised how many folks don't use a product like this, or confuse it with web analytics.
1 reply 0 retweets 1 likeShow this thread -
4. Software developers tend to still think performance is something that happens on the server. It ain't 1999, the web has changed. Nearly always, it's not the server these days - it's the front end. Those web frameworks you love so much are shifting the problem to the user.
1 reply 0 retweets 1 likeShow this thread -
5. You know what's terrible about that? You don't see the problem. So you assume it doesn't exist, while users get really annoyed at your poor performing site.pic.twitter.com/ixWFfr7aB7
1 reply 0 retweets 0 likesShow this thread -
6. "But JD, I use our site, I see the perf!" Sorry, very unlikely you do - you're a software developer who likely has a top 1% computer, likely on a higher grade internet connection, etc. You are not representative of your user base. Use actual data, not your experience.
1 reply 0 retweets 2 likesShow this thread -
7. Here's an example which is actually pretty good compared to what I normally see. Notice that the largest time for this page is in the web-render. It's not the server time. Just like JS Crash Reporting, if you're not tracking this, you likely don't have a clue how bad it is.pic.twitter.com/DRRbXaUEqH
1 reply 0 retweets 2 likesShow this thread -
8. So what's the core cause? All these heavy JS frameworks, bundling everything, too much JS. As cool as you think new-JS framework is, I doubt you're evaluating based on impact on the customer. Too many devs care about new-shiny and not business/user value.
2 replies 0 retweets 2 likesShow this thread -
Replying to @traskjd
If you don’t follow already:
@slightlylate has the metrics, insight, & experiences wrt this. Low-end android devices are horrific, and they’re what the next billion users are running.1 reply 0 retweets 0 likes
They're also what many Americans and Europeans are carrying every day. This isn't so much about geography as it is about class and access.
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.
& Web Standards TL; Blink API OWNER
Named PWAs w/
DMs open. Tweets my own; press@google.com for official comms.