What latencies should i try?
Try 0 and 120 - just anything radically different to throw off the timings.
Just a few comments:
Aptimize pioneered the technique of automatically converting CSS background images into base64 encoded format – this is for both browsers that support data:uri notation and IE6 & IE7 using MHTML.
The Aptimize Website Accelerator product family also support running in the cloud.
Great, thanks Derek. Looks like there are lots of options for sites on shared hosting to automate the optimizations - that’s excellent news.
The connection view is still quite spacey even at 0 ms:
But not spacey at all under 120 ms:
Does anyone have any price quotes on aptimize and / or strangeloop?
Monthly fee? One time fee?
If it is too expensive, I will probably hold off for now. However, it is something I may consider for the future but I would like to have some sort of cost expectation.
This is a little off topic but Is Frankfurt, DE a city in Delaware? When I first looked at it I thought DE stood for denmark
lol, I need to get more consistent about using country codes as more international locations get added. Eventually I’m planning on having the location picker be a map which should help.
I don’t have specific pricing information but it is going to depend on how you use the solution. I’d expect that the cloud-based offerings would be an ongoing fee and the device/software solutions would be an up-front fee with an ongoing maintenance fee.
After my last post in the “start render” topic i ended up finding a link that talked about browser buffering if the charset is not found… Well it just so happens that apparently my dev server wasn’t setup to send the http charset header. I added that and was curious to see if it affected my IE8 results and guess what:
It decreased my total load time on average by 600-700 ms! Thats quite an improvement, all from just adding the charset header. The results are still 100-200 ms slower than the IE7 tests, but that may just be from the specific way i’m maxing out the connections for 2 http connections.
On a side note, the above test shows why i’d recommend deferring the facebook “like” button using something like window.onload. Otherwise the thing is a performance menace, it either times out or takes forever to load.
Wow - very cool. We need to start keeping track of these kernels of knowledge. I should have a wiki page set up soon where we can start to aggregate the knowledge in a central place.