Calculate medians for latest HTTP Archive run

Analysis
1 
SELECT
  NTH(50, quantiles(bytesTotal))   TOTAL_med,
  NTH(50, quantiles(bytesHtmlDoc)) HTMLDOC_med,
  NTH(50, quantiles(bytesHtml))    HTML_med,
  NTH(50, quantiles(bytesJS))      JS_med,
  NTH(50, quantiles(bytesCSS))     CSS_med,
  NTH(50, quantiles(bytesImg))     IMG_med,
  NTH(50, quantiles(bytesGif))     GIF_med,
  NTH(50, quantiles(bytesJpg))     JPG_med,
  NTH(50, quantiles(bytesPng))     PNG_med,
  NTH(50, quantiles(bytesFont))    FONT_med,
  NTH(50, quantiles(bytesFlash))   FLASH_med,
  NTH(50, quantiles(bytesJson))    JSON_med,
  NTH(50, quantiles(bytesOther))   OTHER_med
FROM [httparchive:runs.latest_pages]
1 Like
2

Why do you set the quantiles buckets to 101 vs. 100?

3

@beinwal No particular reason, I think I was just working with a previous example, and it stuck. Updated the query above to use the default (100) bucket count.

4

Mobile stats for July 2013:

22.5KB for HTML + CSS! JavaScript comes in at ~108KB, but some of that (hopefully) is not necessary critical (i.e. can be deferred).

5

Desktop stats for Aug 15 2013:

Median page at ~970KB, with ~50% of that in images. Oh, and it’s a longgg tail (clipped at 7500KB):