How big is your page? (Over 500k?)
TalkieT
Registered Users Posts: 491 Major grins
Ouch... Look at this screenshot of my page load performance... Note the 200+kb .js file... I'm not web savvy enough to know what it is, but it seems large...
Anyone? A crucial part of the page or have I screwed something up by including statcounter and adsense stuff in my page?
http://talkiet.smugmug.com/photos/182357982-O.jpg
And yes, that's showing a 25 second page load time, meassured from Auckland, New Zealand
Anyone? A crucial part of the page or have I screwed something up by including statcounter and adsense stuff in my page?
http://talkiet.smugmug.com/photos/182357982-O.jpg
And yes, that's showing a 25 second page load time, meassured from Auckland, New Zealand
--
http://www.nzsnaps.com (talkiet.smugmug.com)
http://www.nzsnaps.com (talkiet.smugmug.com)
0
Comments
While we're not pleased about long load time (and I promise..double promise we are actively looking into reported speed issues), your graph is a little misleading.
Whatever service you're using to measure performance is omitting 2 very important things: gzip and caching.
Gzip is a compression that takes place on our server before we send the file. For that 200+KB javascript file, you're actually only downloading 47KB.
When we send files, we pass along information to tell the browser to not bother checking for a new file until a certain time. This is caching. This means that the web browser will not even send a request to our web server for certain files. It uses the copies it has on your hard drive.
As you can see from my graph below, the actual amount of data you are downloading is significantly less than what your chart shows. In reality, because of a bug in the way my graphing tool reports cached files you are actually pulling MORE files from your cache than my graph shows (all the .js and .css files are cached).
220KB - 83KB - (78KB <-caching bug) = 59KB <- that is what you are actually downloading
This is NOT how web browsers download things. I think you are getting an invalid picture of what is actually happening in the browser.
Also notice that your chart is reporting 1.2seconds to get your google ads whereas mine reports 115ms (barely over 1/10 of a second). Google would be out of business if it took them 1.2 seconds to serve ads.
I really think that something is odd with your connection or browser.
This first image shows the total requests for the first visit to your site...the browser is pulling everything from our server, nothing from browser cache.
61 requests, 231,773 byte page size in 7.38 seconds
this next image shows what the browser does after it has stored info in cache.
33 requests, 64,533 byte page size in 3.82 seconds.
also notice that no javascript or css is being downloaded.
Thanks Bigwebguy, that's some really interesting analysis there... I had picked up that the service I was using (It's not from my home connection) wasn't doing parallel downloading and that definitely helps a heap... I was also peripherally aware of GZIP compression on files (and certainly the .js files would compress well).
Caching had slipped my mind - I wasn't sure whether the files were dynamically generated or static for good periods of time.
The service I was using was www.websitepulse.com - I setup two tests, one from their server in Auckland, and one from LA. The comparison from LA to Auckland was huge - While NZ was taking 25 seconds (all elements, no compression perhaps, single thread), the LA server was taking 6 seconds with the same constraints.
Over the last 18 odd hours, the average time for the auckland test has been 33.8 seconds, and the LA test - 7.2 seconds.
Since I'm in the middle of an evaluation persiod with their service, if it would be helpful, I'm happy to let you know my login and you can have a look at their data if there's any chance at all it would help.
I should also say that real world performance at this end does seem to be more variable than the testing would indicate, but the bias in real world performance has been for BETTER speed than the test. Right now for example I'm taking about 4 seconds to load a page completely (FF2006, measured from a non-initial load)... This is EXCELLENT performance and it feels really snappy so obviously the test I am using with Websitepulse isn't capturing the real world performance, but it's interesting to compare the difference between US and overseas servers. My local performance however still drags sometimes (up to 20 odd seconds) and it doesn't appear to be related to time of day over here.
Generally though, things are vastly improved from a few weeks ago - Sorry if the graphs from websitepulse alarmed you, but as you point out, they are a theoretical measure and there's many real world considerations that speed up page loading considerably.
Thanks again for your attention though - I learnt a couple of things!
Cheers - N
http://www.nzsnaps.com (talkiet.smugmug.com)
Thanks!