the processing time on the old one is still slow, but the transfer rate is much faster. combining these factors, it’s totally hit or miss as to who wins overall.
The old one is the same as it’s always been. The new one is the only one that should show any difference. We’ll called blog.adrianbischoff.com the ‘control.’
yes, I understand. to clarify my comment: the “old” page has a faster transfer rate than the “new” page. the “new” page does the server side processing faster than the “old” page.
ie, using “time” and “wget” to measure total time and tranfer rate and time, a single sample looks like:
“old”
total time = 5.947s
transfer rate = 337.82 KB/s
transfer time = 0.10905s
“new”
total time = 6.871s
transfer rate = 34.04 KB/s
transfer time = 1.0793s
of course these numbers vary wildly between samples, and I’m not inclined to collect enough data to make it statistically significant…
total time is probably the only data point listed above that you can get a sense for without any measuring tools. and it’s also probably the only one you care about, unless you start hosting lots of big files or something (like a photoblog – ie godhatesmath.com – which incidentally is slow as balls to download the images.) as for “new” vs “old”, I’m taking sample points every 5 minutes and I’ll let you know what it looks like after I’ve gotten a fair number.
I can’t remember if there’s a way to do this with wget or not, but maybe you should try to just get the text too – avid readers (like all of us), will have some of this cached.
re godhatesmath.com: all the text loads quickly, but then the pictures load slowly, filling in line-by-line. we’re in pretty different geo-locations and ip-locations. also this phenomenon goes away once you get the images cached. try to shift+reload them and see. YMMV.
and some stats, based on 235 sample points, sampled every 5 minutes from yesterday afternoon until this morning:
total time (s) transfer rate (KB/s)
mean std mean std
“old” 5.9207 1.4127 280.7420 48.1524
“new” 2.7428 2.1418 70.2967 27.3734
looks like my initial statement was correct. “old” has a much faster transfer rate than “new”, but “new” takes less total time (ie spends less time processing server-side). an interesting thing that I hadn’t noticed before is how much more consistent the total time of “old” is. unfortunately consistently greater total time is probably “worse” than inconsistently lesser total time. looks like overall “new” is probably the “better” choice for this blog.
this geek-rant has been brought to you (like a bunny) by the foundation for better decision making through statistics.
darn you proportional fonts! and darn you wordpress for hating pre! the nbsp version is at least somewhat legible… ok this has been ridiculous. soorrryyyy…
April 17th, 2006 at 2:59 pm
the processing time on the old one is still slow, but the transfer rate is much faster. combining these factors, it’s totally hit or miss as to who wins overall.
April 17th, 2006 at 10:27 pm
The old one is the same as it’s always been. The new one is the only one that should show any difference. We’ll called blog.adrianbischoff.com the ‘control.’
April 18th, 2006 at 6:04 am
yes, I understand. to clarify my comment: the “old” page has a faster transfer rate than the “new” page. the “new” page does the server side processing faster than the “old” page.
ie, using “time” and “wget” to measure total time and tranfer rate and time, a single sample looks like:
“old”
total time = 5.947s
transfer rate = 337.82 KB/s
transfer time = 0.10905s
“new”
total time = 6.871s
transfer rate = 34.04 KB/s
transfer time = 1.0793s
of course these numbers vary wildly between samples, and I’m not inclined to collect enough data to make it statistically significant…
April 18th, 2006 at 9:51 am
hmm. Maybe it’s just wishful thinking, but it seems to me that the new host is universally faster to me.
April 18th, 2006 at 10:35 am
total time is probably the only data point listed above that you can get a sense for without any measuring tools. and it’s also probably the only one you care about, unless you start hosting lots of big files or something (like a photoblog – ie godhatesmath.com – which incidentally is slow as balls to download the images.) as for “new” vs “old”, I’m taking sample points every 5 minutes and I’ll let you know what it looks like after I’ve gotten a fair number.
April 18th, 2006 at 12:06 pm
I can’t remember if there’s a way to do this with wget or not, but maybe you should try to just get the text too – avid readers (like all of us), will have some of this cached.
April 18th, 2006 at 1:51 pm
interesting, I’ve also found godhatesmath.com to be pretty fast.
April 19th, 2006 at 6:15 am
re godhatesmath.com: all the text loads quickly, but then the pictures load slowly, filling in line-by-line. we’re in pretty different geo-locations and ip-locations. also this phenomenon goes away once you get the images cached. try to shift+reload them and see. YMMV.
April 19th, 2006 at 6:57 am
and some stats, based on 235 sample points, sampled every 5 minutes from yesterday afternoon until this morning:
total time (s) transfer rate (KB/s)
mean std mean std
“old” 5.9207 1.4127 280.7420 48.1524
“new” 2.7428 2.1418 70.2967 27.3734
looks like my initial statement was correct. “old” has a much faster transfer rate than “new”, but “new” takes less total time (ie spends less time processing server-side). an interesting thing that I hadn’t noticed before is how much more consistent the total time of “old” is. unfortunately consistently greater total time is probably “worse” than inconsistently lesser total time. looks like overall “new” is probably the “better” choice for this blog.
this geek-rant has been brought to you (like a bunny) by the foundation for better decision making through statistics.
April 19th, 2006 at 7:05 am
blarg, that table came out like crap. who knows what wordpress allows. here’s an attempt with a pre tag:
total time (s) transfer rate (KB/s)
mean std mean std
“old” 5.9207 1.4127 280.7420 48.1524
“new” 2.7428 2.1418 70.2967 27.3734
and an attempt with non-breaking spaces:
total time (s) transfer rate (KB/s)
mean std mean std
“old”†5.9207 1.4127 280.7420 48.1524
“new” 2.7428 2.1418 70.2967 27.3734
April 19th, 2006 at 7:10 am
darn you proportional fonts! and darn you wordpress for hating pre! the nbsp version is at least somewhat legible… ok this has been ridiculous. soorrryyyy…
April 19th, 2006 at 9:29 am
That’s pretty interesting about the times and confirms my thoughts.
As for godhatesmath, even ctrl-refreshing it loads pretty fast (each page