community.centminmod.com journey for speed

Revisiting this forum's index page Webpagetest across a variety of browser clients and devices for 5Mbps cable and 4G mobile speeds (9Mbps) for following:

• Dulles, VA - Chrome - Cable
• Dulles, VA - Thinkpad T430 - Chrome - Cable
• Dulles, VA - Asus Aspire i3 2017 - Chrome - Cable
• Dulles, VA - Raspberry Pi 3 B - Chromium - Cable
• Dulles, VA - Raspberry Pi 4 - Chromium - Cable
• Dulles, VA - Moto G4 - Chrome - 4G
• Dulles, VA - Alcatel 1X - Chrome - 4G
• Dulles, VA - OnePlus 5 - Chrome - 4G

---

 

Webpagetest comparison tests continued from previous post for following browser client speed profiles

• Dulles, VA - Chrome - Cable
• Dulles, VA - Thinkpad T430 - Chrome - Cable
• Dulles, VA - Asus Aspire i3 2017 - Chrome - Cable
• Dulles, VA - Raspberry Pi 3 B - Chromium - Cable
• Dulles, VA - Raspberry Pi 4 - Chromium - Cable
• Dulles, VA - Moto G4 - Chrome - 4G
• Dulles, VA - Alcatel 1X - Chrome - 4G
• Dulles, VA - OnePlus 5 - Chrome - 4G

 

Google Lighthouse v6 and PageSpeed Insights v6 have been released with updated metrics outlined at WebPerf - PageSpeed - How to use webpagetest.org for page load speed testing. This is the first month after release of the new metrics so will track how this forum's main domain real world user page speed fairs via Google Chrome User Experience Reports (CRuX) data.

Bear in mind this is real world users so if more slow visitors visit the forums and domain, then my averages would be lower even if I have improved my pagespeed optimisations. For example, if a visitor is watching a video on his PC, that can slow web browser render/load times dramatically - my test showed page render time for LCP jumped from 1-2 seconds to 19+ seconds WebPerf - PageSpeed - How to use webpagetest.org for page load speed testing !

First my Google Core Web Vital metrics for April 2020



Then my Google Core Web Vital metrics for May 2020



Then historic metrics for past ~9 months as you can see some metrics are now slower - probably due to my increased visitor traffic from slow countries in Asia i.e. India, China and from South America, Brazil which would bring down my averages + changes in how many Google Adsense ads I have no on some pages being more than in the past.





You can get these reports from Google CRuX dashboard reports via Google Data Studio http://g.co/chromeuxdash

 

My Google Datastudio's custom Google Analytics sourced data dashboards tracking page speed for top visitor countries. As you can see China, India, Indonesia, Brazil and even a spike from Germany have pulled down my pagespeed averages compared to USA !

Overall for Jan 1 to Jun 10, 2020


USA vs India


USA only

India only

China only

Germany only - wow 200+ second spike at end of April 2020 !


Indonesia only with spike to 40 seconds in Jan 2020



Brazil only

 

Not satisfied with default Google CRuX Reports, I created my own via Google Data Studio with combined Google Analytics + Google CRuX data sources - an insight into my year to date CRuX LCP, FCP, TTB, FID metrics as well as month to date GA metrics

 

Finally got both mobile and desktop page speed metrics optimised into the 'good URLs' category for Google Webmaster Console's Speed Report for real world LCP and FID metrics

 

Google Core Web Vital metrics can be exposed in Google Chrome User Experience (CRuX) report for real world visitors metrics. Previously posted my April vs May 2020 Core Web Vital metric numbers here. CRuX is updated monthly every 2nd Tuesday of the month. So here's June 2020 metrics

First my custom Google DataStudio Core Web Vital dashboard with June 2020 metrics added for select metrics which averages both mobile + desktop.

It focuses on LCP, FCP, TTFB and FID broken down by monthly basis. I see a dramatic improvement since I started re-optimizing my forums for FCP and thus LCP and improving my caching for better TTFB. To optimize for LCP, you need to optimize for FCP which also needs fast TTFB

• April Good LCP = 74.83% vs May Good LCP = 78.42% vs June Good LCP = 88.72% which is 13.13% improvement over previous May
• April Good FCP = 57.53% vs May Good FCP = 56.3% vs June Good FCP = 71.55% which is 27.09% improvement over previous May
• April Good TTFB = 32.52% vs May Good TTFB = 30.32% vs June Good TTFB = 49.79% which is 64.2% improvement over previous May. I believe the TTFB improvement came from re-enabling Cloudflare Argo which improves one aspect of pagespeed = server connection response time.

Cloudflare Argo Analytics with past 48hrs showing ~86.7% of CF requests being Argo smart routed with an average 63.31% improvement in Cloudflare edge server response time from 447ms to 164ms across all 200+ Cloudflare datacenters.



Cloudflare Argo Analytics via Cloudflare API with custom script I wrote to breakdown Argo performance on a per Cloudflare datacenter basis.

Example largest Argo improvement (max) came from MCI datacenter (IATA airport code for Kansas City International Airport) with 71.83% improvement using CF Argo reducing average CF edge server response time 389ms to 109ms. Least Argo improvement (min) came from LAX datacenter (IATA airport code for Los Angeles International Airport) with 16.82% Argo improvement - reducing average CF edge server response time from 69ms to 57ms. LAX was a small improvement as my origin server is in US West Coast so already close to that regions users.

Code (Text):

./cf-analytics.sh 24hrs
------------------------------------------------------------------
  Cloudflare Argo Analytics
------------------------------------------------------------------
  since: 2020-07-12T08:05:00Z
  until: 2020-07-14T08:05:00Z
------------------------------------------------------------------
  Argo Response Time:
------------------------------------------------------------------
  request-without-argo: 50041
  request-with-argo: 451927
  argo-smarted-routed: 86.7%
  argo-improvement: 63.3200%
  without-argo: 447 (milliseconds)
  with-argo 164 (milliseconds)
------------------------------------------------------------------
  Argo Cloudflare Datacenter Response Times
------------------------------------------------------------------
  IAD  23544  0.5924863260953607   259.92594675665543  105.92337750594632
  DFW  18157  0.618428917655495    281.07032057911067  107.24830643828827
  FRA  16151  0.5988762547459494   223.76420454545453  89.75713578106618
  AMS  14615  0.6093226069362514   342.6161330822348   133.85237769414985
  SIN  12737  0.6557943167116163   542.7400295420975   186.81420271649526
  LHR  12390  0.6004527460647228   263.0115758028379   105.08555286521388
  PRG  12080  0.6386576112954431   166.53088803088804  60.174668874172184
  HKG  11765  0.6846424762744965   646.9771494965144   204.02911177220568
  BOM  10911  0.6644424611209188   678.7832225913621   227.7708276051691
  GRU  10875  0.6660538120444008   407.94995794785535  136.23333333333332
  SYD  10709  0.6598913519441224   648.2338308457712   220.46993183303763
  SJC  10411  0.6600332224724011   264.51914311759344  89.92772068004994
  MIA  10403  0.5634598253456292   117.1458141674333   51.13885417667981
  MXP  10318  0.5429896209433198   87.49567474048443   39.98643147896879
  CMN  10221  0.6192282009411394   190.93591905564924  72.70301340377654
  KWI  10217  0.640950830787479    539.6357012750456   193.7557502202212
  LOS  10210  0.6411994906758769   329.17109929078015  118.10675808031341
  JIB  10190  0.631719538686767    505.34965034965035  186.11040235525024
  DOH  10188  0.6311527891724791   390.59928443649375  144.07145661562623
  MDE  10182  0.5518766821234001   277.9855195911414   124.57179336083284
  PTY  10171  0.6062378210101277   477.9572907679034   188.2015042768656
  TPE  10166  0.7137341629895922   766.4925023430178   219.42061774542594
  MCT  10166  0.5981826987721713   598.4883720930233   240.48298249065513
  ATL  10166  0.6238817365726111   283.1781115879828   106.50845957111942
  LAD  10162  0.4132042385305727   388.65215479331573  228.05943711867744
  LIM  10160  0.6893435153201566   581.3349514563107   180.5954724409449
  BKK  10147  0.6311335263703575   610.7763401109057   225.294914753129
  KIX  10145  0.6643034103584737   735.5272727272727   246.91399704287826
  KUL  10134  0.6466351007402438   717.9140127388536   253.68561278863234
  DEL  10127  0.5674534466035991   463.73801220575416  200.5882788584971
  JNB  10125  0.6337697589422374   497.1270598438855   182.06296296296296
  MBA  10122  0.6066559415913731   605.8981001727116   238.32641770401108
  EZE  10101  0.6840626666540829   511.10243055555554  161.47633897633898
  SCL  10093  0.645800685773324    605.7266217354676   214.54795402754385
  MAA  10084  0.64929606424822     715.979381443299    251.09678698928997
  MEL  10043  0.6554878172559994   881.8089788732394   303.79393607487805
  PER  10035  0.6468417836090606   790.1743159752868   279.0565520677628
  SEA  7685   0.34263549472718496  141.14790286975716  92.78562134027327
  ARN  4267   0.6464342747009824   614.8333333333334   217.38399343801265
  LAX  3946   0.1682495010815132   69.0033407572383    57.393563101875316
  CDG  2445   0.5837100728147353   513.6693548387098   213.8353783231084
  EWR  1990   0.6587059760300981   315.3310502283105   107.62060301507537
  YUL  1927   0.6700574081564253   368.92533936651586  121.72418266735859
  DME  1695   0.5919418117174459   666.4306358381502   271.9424778761062
  BNE  1059   0.5341457740467057   580.3861788617886   270.3753541076487
  ORD  891    0.6462723899855612   298.24257425742576  105.496632996633
  NRT  863    0.5343615692118526   340.8333333333333   158.70509849362688
  MUC  834    0.6527204222975816   616.6463414634146   214.14868105515586
  WAW  817    0.580703894118265    524.3604651162791   219.86230110159119
  ICN  799    0.6018882692707036   436.59638554216866  173.81414267834793
  MCI  705    0.7183330298488647   389.4871794871795   109.70567375886525
  MAN  652    0.6979107114852449   645.9868421052631   195.1457055214724
  DUS  618    0.6811820684027576   630.5909090909091   201.04368932038835
  HEL  605    0.6461083480126698   603.8291139240506   213.6900826446281
  KBP  516    0.5576086068682248   597.9               264.50581395348837
  ATH  512    0.6764601572632944   710.1315789473684   229.755859375
  SOF  419    0.5395089524601031   578.2843137254902   266.2947494033413
  MAD  419    0.5956105054456735   510.22727272727275  206.33054892601433
  total-argo-reqs: 452085
  datacenter-calc-avg-resp-without: 447.6710
  datacenter-calc-avg-resp-with: 164.3573
  argo-improvement:
      min: 0.1682
      avg: 0.6116
      max: 0.7183
      stddev: 0.0875
  argo-resp-time-without-argo:
      min: 69.0033
      avg: 474.9855
      max: 881.8090
      stddev: 194.8987
  argo-resp-time-with-argo:
      min: 39.9864
      avg: 176.3690
      max: 303.7939
      stddev: 66.9147

Then CRuX June 2020 reports averaging both mobile + desktop for 75% percentile of real world visitors (using Chrome browser) with LCP for June improving it's 75% metric from 2.2s to 1.6s dropping 0.6s


LCP


FCP



TTFB

 

Previous post was for Google Core Web Vital metrics stats for July 2020. Here's an update and overview for August to October 2020

First my custom Google DataStudio Core Web Vital dashboard with August to October 2020 metrics added for select metrics which averages both mobile + desktop.


Then Google Chrome User Experience (CRuX) report for real world visitors metrics


And my Google Webmaster Tools Core Web Vitals

 

With the growing importance of Google's Core Web Vital metrics, I've started to track TTFB, CLS, LCP, FCP and FID metrics via custom Google Analytics reporting via Google Web Vitals library. Only difference if I am tracking all web browsers and not just Google Chrome

Browser console logged metrics



Google Analytics saved reports for Core Web Vital metrics



CLS metrics need whole integers so the value is multiplied by 1000x so reported value divide by 1000 is the CLS value



LCP



TTFB



Custom Google Analytics Dashboard charts