Current time: 12-14-2017, 08:30 AM Hello There, Guest! (LoginRegister)

Post Reply 
 
Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Revisiting browser and connectivity defaults
03-17-2013, 06:50 AM
Post: #21
RE: Revisiting browser and connectivity defaults
Yep, just your luck. One of the 35 test machines (newly imaged for the migration) wasn't traffic shaping correctly and you happened to catch that machine. All fixed and I verified all of the machines so it shouldn't happen again.
Visit this user's website Find all posts by this user
Quote this message in a reply
03-17-2013, 12:14 PM
Post: #22
RE: Revisiting browser and connectivity defaults
Sweet, thank you Patrick. When I first saw the results I was like "Yeah, they finally pared down the content!" and then. Sad LOL We're going to use a Google graph to put the graphs up on our monitors around work as a "Wall of pain" for our customers. Would have been Wall of Shame, but that was already taken by something else. Smile
Find all posts by this user
Quote this message in a reply
03-18-2013, 10:52 AM
Post: #23
RE: Revisiting browser and connectivity defaults
Interesting that Akamai's state of the Internet says the avg speed in the US is 7 Mbps, and Netflix says it's 2.3 Mbps. The truth is probably in between. So, Cable 5M is good, I agree.

Question about latency: Why simulate latency, doesn't that naturally occur from the distance of the test? I'd be inclined to vote for 0ms latency by default.
Find all posts by this user
Quote this message in a reply
03-19-2013, 12:54 AM
Post: #24
RE: Revisiting browser and connectivity defaults
(03-18-2013 10:52 AM)koop Wrote:  Interesting that Akamai's state of the Internet says the avg speed in the US is 7 Mbps, and Netflix says it's 2.3 Mbps. The truth is probably in between. So, Cable 5M is good, I agree.

Question about latency: Why simulate latency, doesn't that naturally occur from the distance of the test? I'd be inclined to vote for 0ms latency by default.

Averages are evil. High-end connections could easily skew the average way high so I prefer to look at percentiles. The Akamai report has percent of adoption for 4Mbps which was the more interesting piece of data in the report and that comes out at 62% for the US if I'm remembering correctly. That means that 62% of the Internet users in the US have a connection that is at least 4Mbps as measured by Akamai.

WebPagetest traffic shapes the last-mile of a connection only. All tests still have the added latency of the actual geography and routing but different last-mile technologies have different amounts of latency that they add. The 28ms number for cable comes from the 2011 FCC broadband report where they measured the latencies within ISP networks to end users (2012 report numbers are pretty close) and it lines up with other data I have seen as well. Using a 0ms latency for last-mile would give everyone the awesome latencies you see from fiber which isn't particularly realistic.
Visit this user's website Find all posts by this user
Quote this message in a reply
03-19-2013, 01:06 AM (This post was last modified: 03-19-2013 01:09 AM by pmeenan.)
Post: #25
RE: Revisiting browser and connectivity defaults
(03-17-2013 12:14 PM)mle_ii Wrote:  Sweet, thank you Patrick. When I first saw the results I was like "Yeah, they finally pared down the content!" and then. Sad LOL We're going to use a Google graph to put the graphs up on our monitors around work as a "Wall of pain" for our customers. Would have been Wall of Shame, but that was already taken by something else. Smile

Have you considered looking at an automation solution to help? There are several free and commercial offerings that could do most of the grunt work.

This is roughly what mod_pagespeed (free/open-source) could do:

Cable: http://www.webpagetest.org/result/130318...0a1b11b96/

DSL: http://www.webpagetest.org/result/130318...65d611a6b/
Visit this user's website Find all posts by this user
Quote this message in a reply
03-19-2013, 02:51 AM
Post: #26
RE: Revisiting browser and connectivity defaults
(03-19-2013 12:54 AM)pmeenan Wrote:  WebPagetest traffic shapes the last-mile of a connection only. All tests still have the added latency of the actual geography and routing but different last-mile technologies have different amounts of latency that they add.

Does it use max(configured latency, actual latency)? Or does it simply add "configured latency" to every round trip?
Find all posts by this user
Quote this message in a reply
03-20-2013, 01:42 AM
Post: #27
RE: Revisiting browser and connectivity defaults
It adds the "configured latency" to every round trip since it's simulating that amount of latency for the last-mile of a leg.
Visit this user's website Find all posts by this user
Quote this message in a reply
Post Reply 


Forum Jump:


User(s) browsing this thread: 1 Guest(s)