Hi,

I recently tried Maxcdn, and following some problems I experienced with them, due to a bad configuration I made, but lacking support to fix it, I tried Edgecast services through Speedyrails.

Basically, I tested the same page, the homepage of my website at www.webpagetest.org when the website was served by Maxcdn, Edgecast, and my regular shared hosting account. I tested 3 location for each situation.

I'm not sure this is an optimal test since I did not test the results over time. I ran only one instance of www.webpagetest.org for each location to see how each configuration performed. I'd thought I'd share those anyway, but feel free to tell me if these are any good. :)

First number is first view, second is repeat view.

Paris :
No CDN : (15.092s) (14.500s)
MaxCDN : (10.302s)(6.004s)
Edgecast : (10.960s) (2.768s)

Singapore :
No CDN : (8.422s) (7.190s)
MaxCDN : (8.834s) (3.838s)
Edgecast : (17.524s) (8.351s)

NYC :
No CDN : (4.989s) (4.298s)
MaxCDN : (4.370s) (1.461s)
Edgecast : (5.551s) (1.529s)

To be honest, the only conclusion I can draw is that obviously, CDN adds in performance. But as far as the comparison between MaxCDN and Edgecast is concerned, I dont know.

What are you concerned of between the two? It's common that certain location is faster for certain CDNs.

You should probably run Webpagetest twice so that you know you're seeing a cache hit result. Otherwise, the time may be slower as the cdn has to go grab the file from your server and then deliver it. Try running 5 or 7 tests and using the median or average.

I'll try that, I'll keep you guys posted.

However here is what bothers me right now. When the website is served by Edgecast, and following a webpagetest.org test with Paris as test area, the waterfall shows the route to go back to LA. Please see attachment.

I'm no pro, but this is abnormal right?

Edit : sorry, wrong attachment, see next post

here is the screenshot. And it's the same thing for Maxcdn btw. What's the point in having a CDN if the content is pulled from a server 8000 miles away from the user??

It's not actually serving the files from California - that's just the IP that you're seeing on their anycast network.

first of all thank you very much to give us this kind of statistic would you try with justping those companies i found this site and i found very handy

http://www.just-ping.com/

for example cdn.1231.edgecast.com

thanks

Hi,

I recently tried Maxcdn, and following some problems I experienced with them, due to a bad configuration I made, but lacking support to fix it, I tried Edgecast services through Speedyrails.

Basically, I tested the same page, the homepage of my website at www.webpagetest.org when the website was served by Maxcdn, Edgecast, and my regular shared hosting account. I tested 3 location for each situation.

I'm not sure this is an optimal test since I did not test the results over time. I ran only one instance of www.webpagetest.org for each location to see how each configuration performed. I'd thought I'd share those anyway, but feel free to tell me if these are any good. :)

First number is first view, second is repeat view.

Paris :
No CDN : (15.092s) (14.500s)
MaxCDN : (10.302s)(6.004s)
Edgecast : (10.960s) (2.768s)

Singapore :
No CDN : (8.422s) (7.190s)
MaxCDN : (8.834s) (3.838s)
Edgecast : (17.524s) (8.351s)

NYC :
No CDN : (4.989s) (4.298s)
MaxCDN : (4.370s) (1.461s)
Edgecast : (5.551s) (1.529s)

To be honest, the only conclusion I can draw is that obviously, CDN adds in performance. But as far as the comparison between MaxCDN and Edgecast is concerned, I dont know.

you are correct i test also many companies except maxcdn and you can tell another then akamai they all connect us server what is the point user comes from japan to connect us espcially company has edge server in tokyo japan believe me akamai also connect japan but not very good as good as i though, but apart asia region you must use cdn not for speed you need to use for reducing cpu you won't save any bandwith to use cdn at all same picture png jpg same data. only change cdn price bandwith start from $0.20 to $0.60 as you can see wht also use cdn even they have plenty server db etc. you need to distribute load many reliable servers.



here is the screenshot. And it's the same thing for Maxcdn btw. What's the point in having a CDN if the content is pulled from a server 8000 miles away from the user??

It's not actually serving the files from California - that's just the IP that you're seeing on their anycast network.

You're right, here is the answer given by Edgecast :

The 93.184.220.20 IP is our global load balancing IP which is not assigned to a specific EdgeCast POP. The customer would have to do a traceroute to wpc.XXX.edgecastcdn.net (where XXXX is their customer ID/account number) or 93.184.220.20. If they are based in the United Kingdom, they should be routed to London and it should be noted in the traceroute result.

first of all thank you very much to give us this kind of statistic would you try with justping those companies i found this site and i found very handy

http://www.just-ping.com/

for example cdn.1231.edgecast.com

thanks

That's a good idea. For the url Edgecast gave me :

Singapore, Singapore: Okay 2.0 2.3 2.6 68.232.44.19
Amsterdam2, Netherlands: Okay 0.8 0.9 1.0 93.184.220.20
Florida, U.S.A.: Okay 40.9 41.1 41.5 72.21.91.19
Amsterdam3, Netherlands: Okay 0.5 0.6 0.6 93.184.220.20
Hong Kong, China: Okay 56.7 60.5 68.6 68.232.44.19
Sydney, Australia: Okay 0.5 1.6 6.9 68.232.44.19
München, Germany: Okay 8.2 8.4 8.5 93.184.220.20
Cologne, Germany: Okay 4.5 4.6 4.9 93.184.220.20
New York, U.S.A.: Okay 9.9 10.7 13.2 72.21.91.19
Cairo, Egypt: Okay 203.4 205.0 207.5 93.184.220.20
Amsterdam, Netherlands: Okay 0.5 0.5 0.5 93.184.220.20
Stockholm, Sweden: Okay 28.1 28.7 29.2 93.184.220.20
Santa Clara, U.S.A.: Okay 4.2 4.6 5.1 72.21.91.19
Vancouver, Canada: Okay 4.8 4.9 5.2 72.21.91.19
Krakow, Poland: Okay 30.5 30.6 31.0 93.184.220.20
London, United Kingdom: Okay 0.9 1.3 1.6 93.184.220.20
Madrid, Spain: Okay 17.4 17.6 17.7 93.184.220.20
Padova, Italy: Okay 28.8 29.0 29.3 93.184.220.20
Austin, U.S.A.: Okay 29.8 29.9 30.2 72.21.91.19
Amsterdam, Netherlands: Okay 0.4 0.5 0.9 93.184.220.20
Paris, France: Okay 0.4 0.6 0.8 93.184.220.20
Melbourne, Australia: Okay 15.5 16.5 17.2 68.232.44.19
Shanghai, China: Okay 194.0 195.1 196.4 68.232.44.19
Copenhagen, Denmark: Okay 16.0 16.5 17.3 93.184.220.20
Lille, France: Okay 7.7 7.8 8.1 93.184.220.20
San Francisco, U.S.A.: Okay 2.8 3.3 3.8 72.21.91.19
Zurich, Switzerland: Okay 13.9 14.2 14.6 93.184.220.20
Mumbai, India: Okay 244.6 259.7 311.4 68.232.44.19
Chicago, U.S.A.: Okay 0.3 0.4 0.8 72.21.91.19
Nagano, Japan: Okay 114.0 114.1 114.4 68.232.44.19
Haifa, Israel: Okay 65.5 66.1 68.5 93.184.220.20
Auckland, New Zealand: Okay 129.5 129.7 129.9 68.232.44.19
Antwerp, Belgium: Okay 5.2 5.3 5.5 93.184.220.20
Groningen, Netherlands: Okay 4.2 4.7 5.1 93.184.220.20
Moscow, Russia: Okay 46.6 46.6 46.8 93.184.220.20
Dublin, Ireland: Okay 13.5 13.8 14.2 93.184.220.20
Oslo, Norway: Okay 24.0 24.4 28.0 93.184.220.20
Kharkov, Ukraine: Okay 42.6 42.7 42.9 93.184.220.20
Manchester, United Kingdom: Okay 7.5 12.6 47.6 93.184.220.20
Vilnius, Lithuania: Okay 30.0 30.2 30.5 93.184.220.20
Ashburn, U.S.A.: Okay 3.8 4.0 4.1 72.21.91.19
Bucharest, Romania: Okay 35.2 35.3 35.4 93.184.220.20
Bangkok, Thailand: Okay 29.2 29.4 30.3 68.232.44.19
Kuala Lumpur, Malaysia: Packets lost (10%) 180.8 181.0 181.5 68.232.44.19
Jakarta, Indonesia: Okay 14.3 15.1 20.8 68.232.44.19
Cape Town, South Africa: Okay 150.3 150.4 150.7 93.184.220.20
Glasgow, United Kingdom: Okay 11.3 11.4 11.7 93.184.220.20
Lisbon, Portugal: Okay 34.3 35.8 37.4 93.184.220.20
Chicago, U.S.A.: Okay 1.5 1.6 2.1 72.21.91.19
Dallas, U.S.A.: Okay 1.4 1.5 1.7 72.21.91.19
Buenos Aires, Argentina: Okay 162.3 162.4 162.6 72.21.91.19
Istanbul, Turkey: Okay 47.1 47.3 47.9 93.184.220.20
Gdansk, Poland: Okay 29.7 52.1 100.0 93.184.220.20
Beijing, China: Okay 202.4 208.2 214.6 72.21.91.19

And for MaxCDN :

Singapore, Singapore: Okay 185.5 186.0 186.8 67.201.31.38
Amsterdam2, Netherlands: Okay 1.2 1.2 1.4 67.201.31.38
Florida, U.S.A.: Okay 33.6 33.8 34.1 67.201.31.38
Amsterdam3, Netherlands: Okay 0.9 1.0 1.2 67.201.31.38
Hong Kong, China: Okay 167.6 168.2 169.1 67.201.31.38
Sydney, Australia: Okay 165.1 166.1 169.4 67.201.31.38
München, Germany: Okay 21.6 21.9 22.1 67.201.31.38
Cologne, Germany: Okay 19.4 21.2 26.4 67.201.31.38
New York, U.S.A.: Okay 4.3 4.5 4.5 67.201.31.38
Cairo, Egypt: Okay 135.5 136.9 141.5 67.201.31.38
Amsterdam, Netherlands: Okay 0.8 0.9 1.1 67.201.31.38
Stockholm, Sweden: Packets lost (10%) 40.9 41.7 42.2 67.201.31.38
Santa Clara, U.S.A.: Okay 16.5 16.7 17.1 67.201.31.38
Vancouver, Canada: Okay 186.2 186.3 186.5 67.201.31.38
Krakow, Poland: Okay 32.7 33.0 33.2 67.201.31.38
London, United Kingdom: Okay 10.0 10.4 10.7 67.201.31.38
Madrid, Spain: Okay 39.8 40.0 40.2 67.201.31.38
Padova, Italy: Okay 37.6 37.8 38.1 67.201.31.38
Austin, U.S.A.: Okay 134.1 134.3 134.5 67.201.31.38
Amsterdam, Netherlands: Okay 0.9 1.0 1.1 67.201.31.38
Paris, France: Okay 17.8 18.0 18.4 67.201.31.38
Melbourne, Australia: Okay 179.0 179.8 181.1 67.201.31.38
Shanghai, China: Okay 169.6 170.1 170.8 67.201.31.38
Copenhagen, Denmark: Okay 14.0 14.6 15.2 67.201.31.38
Lille, France: Okay 18.6 19.1 19.5 67.201.31.38
San Francisco, U.S.A.: Okay 11.1 11.6 12.1 67.201.31.38
Zurich, Switzerland: Okay 40.9 41.0 41.1 67.201.31.38
Mumbai, India: Okay 144.8 151.0 173.3 67.201.31.38
Chicago, U.S.A.: Okay 23.9 24.0 24.1 67.201.31.38
Nagano, Japan: Okay 117.5 118.2 118.9 67.201.31.38
Haifa, Israel: Okay 77.1 77.4 77.6 67.201.31.38
Auckland, New Zealand: Okay 131.6 131.8 132.4 67.201.31.38
Antwerp, Belgium: Okay 5.3 5.5 5.6 67.201.31.38
Groningen, Netherlands: Okay 4.8 5.4 6.0 67.201.31.38
Moscow, Russia: Okay 46.7 46.7 46.7 67.201.31.38
Dublin, Ireland: Okay 20.0 20.4 20.6 67.201.31.38
Oslo, Norway: Unknown host: papaye.papayeprod.netdna-cdn.com
Kharkov, Ukraine: Okay 116.6 116.8 116.9 67.201.31.38
Manchester, United Kingdom: Okay 16.6 16.7 16.8 67.201.31.38
Vilnius, Lithuania: Okay 47.4 47.7 47.9 67.201.31.38
Ashburn, U.S.A.: Okay 2.5 2.6 3.0 67.201.31.38
Bucharest, Romania: Okay 47.1 47.2 47.4 67.201.31.38
Bangkok, Thailand: Okay 209.5 209.7 210.1 67.201.31.38
Kuala Lumpur, Malaysia: Okay 182.0 182.2 183.0 67.201.31.38
Jakarta, Indonesia: Okay 231.2 232.3 239.6 67.201.31.38
Cape Town, South Africa: Okay 159.4 159.6 159.7 67.201.31.38
Glasgow, United Kingdom: Okay 20.0 20.3 21.2 67.201.31.38
Lisbon, Portugal: Okay 44.7 44.8 44.9 67.201.31.38
Chicago, U.S.A.: Okay 24.8 25.0 25.3 67.201.31.38
Dallas, U.S.A.: Okay 21.4 21.4 21.5 67.201.31.38
Buenos Aires, Argentina: Okay 158.2 158.3 158.4 67.201.31.38
Istanbul, Turkey: Okay 49.2 49.4 49.7 67.201.31.38
Gdansk, Poland: Okay 34.8 48.9 100.0 67.201.31.38
Beijing, China: Okay 186.1 315.3 473.8 67.201.31.38

The Edgecast results do show some local pings whereas Maxcdn's dont.

However, I'm experiencing something I dont understand. Yesterday I switched back to Maxcdn to test them again. I've implementing their url in the W3 Total Cache plugin. This url shows in my source code. However, when I test it on webpagetest.org, I still get the Egdecast results (with Edgecast url). Do you know where that comes from?

Did you empty the page cache?

Yes I emptied all caches throught W3 total cache.

yes because webpagetest.org has dns cache time thats why still lookat old pictures, thats normal try after 6 hours, my understanding is whether they show ip US but they absolutely connect local edge server i dont know how do they do it as you can see (latency) i cannot ping any US server from sydney less then 45ms impossible at this technology but i can see edge ip from US but it connect sydney edge server latency is less then 45


You're right, here is the answer given
However, I'm experiencing something I dont understand. Yesterday I switched back to Maxcdn to test them again. I've implementing their url in the W3 Total Cache plugin. This url shows in my source code. However, when I test it on webpagetest.org, I still get the Egdecast results (with Edgecast url). Do you know where that comes from?

yes because webpagetest.org has dns cache time thats why still lookat old pictures, thats normal try after 6 hours

Good point. It didn't even occur to me that he'd be making the changes at the DNS level. When I'm testing something like CDN's, I'll just set up separate CNAME records for each network. That way I can switch back and forth between them immediately.

yes because webpagetest.org has dns cache time thats why still lookat old pictures, thats normal try after 6 hours, my understanding is whether they show ip US but they absolutely connect local edge server i dont know how do they do it as you can see (latency) i cannot ping any US server from sydney less then 45ms impossible at this technology but i can see edge ip from US but it connect sydney edge server latency is less then 45
I did not understand everything you said.

Do you mean anycast (http://en.wikipedia.org/wiki/Anycast)?