Google
 
Web www.howardforums.com
Pages: 1

Hard coded proxy

(Click here to view the original thread with full colors/images)

Posted by: prubin

With my 8525 (now with the upgraded ROM), what are the advantages and disadvantages to disabling the proxy in Internet Explorer?
I have experimented today with both on & off. I did not notice anything other than a speed slow down (3g area) with proxy disabled.
Thanks-


Posted by: wpbear

Quote:
Originally Posted by prubin
With my 8525 (now with the upgraded ROM), what are the advantages and disadvantages to disabling the proxy in Internet Explorer?
I have experimented today with both on & off. I did not notice anything other than a speed slow down (3g area) with proxy disabled.
Thanks-

If you permanently disable the proxy:

No More CV

No More MediaNet Homepage

Probably buying ringtones etc from Cingular Site will be screwed

Your IE on the device will work when connected through activesync using the internet from your PC

Some VPN connections have a hard time with the proxy enabled.

Some will argue speed, I do not notice a difference either way.


Posted by: RF9

The 3G speed is not actually slower with the proxy disabled. If you're running a bandwidth test like http://www.dslreports.com/mspeed then here's what's going on:

With the proxy enabled:
The speed test downloads a image (.GIF or .JPG) to your phone. It times how long it takes for your phone to download it and caclulates your bandwidth. The proxy reduces images to smaller sizes to that they'll fit better on a mobile browser. Therefore the 1MB image is reduced to something like 700kb. DSL reports calculages how long it took to download 1MB when you only downloaded 700kb.
Therefore the test is "tricked" or thrown off to give you a flase high number.

With the proxy disbaled:
You're running the test without image manipulation and the test is accurate.

You'll see lot of Treo 750 owners reporting 900kb or more with HSDPA hack enbaled or 400kb with UMTS only. you can't believe ANY of them because all Cingular Troe 750's have the proxy hard coded as well.

The biggest down side to leaving the proxy enabled is degraded image quality. The proxy will shrink images and remove detail. So when I download real time traffic maps, weather radar images, or really large images (like a newspaper ad from the mercury news web site) the image is so scrunched that I can't make out any of the detail on it.

Leaving the proxy on may result in faster web page downloads because it reduces the amount of data you're downloading. It also lets you continue to use media net and cingular video sites.
However the proxy may result in increased latency (the time before the web page stars downlooading.) but this latency is usually very short and the end result could be faster web page downloads overall.

Here's my web page on the subject (for Treo 750 owners)
http://www.gadgetech.info/treo/hacks/disableproxy/
My thread from last month on this subject:
http://www.howardforums.com/showthread.php?p=8555248


Posted by: tbl178

good info. thanks rf9


Posted by: prubin

much appreciated - for the time being I will leave it enabled.


Posted by: michaelflorio82

Quote:
Originally Posted by RF9
The 3G speed is not actually slower with the proxy disabled. If you're running a bandwidth test like http://www.dslreports.com/mspeed then here's what's going on:

With the proxy enabled:
The speed test downloads a image (.GIF or .JPG) to your phone. It times how long it takes for your phone to download it and caclulates your bandwidth. The proxy reduces images to smaller sizes to that they'll fit better on a mobile browser. Therefore the 1MB image is reduced to something like 700kb. DSL reports calculages how long it took to download 1MB when you only downloaded 700kb.
Therefore the test is "tricked" or thrown off to give you a flase high number.

With the proxy disbaled:
You're running the test without image manipulation and the test is accurate.

You'll see lot of Treo 750 owners reporting 900kb or more with HSDPA hack enbaled or 400kb with UMTS only. you can't believe ANY of them because all Cingular Troe 750's have the proxy hard coded as well.

The biggest down side to leaving the proxy enabled is degraded image quality. The proxy will shrink images and remove detail. So when I download real time traffic maps, weather radar images, or really large images (like a newspaper ad from the mercury news web site) the image is so scrunched that I can't make out any of the detail on it.

Leaving the proxy on may result in faster web page downloads because it reduces the amount of data you're downloading. It also lets you continue to use media net and cingular video sites.
However the proxy may result in increased latency (the time before the web page stars downlooading.) but this latency is usually very short and the end result could be faster web page downloads overall.

Here's my web page on the subject (for Treo 750 owners)
http://www.gadgetech.info/treo/hacks/disableproxy/
My thread from last month on this subject:
http://www.howardforums.com/showthread.php?p=8555248


Just a question for you RF. In another thread I had posted pretty much what you said about img compression, and someone ekse posted this? Can you give me some insight to which one is correct?

Quote:
Originally Posted by michaelflorio82
Thats correct, except the proxy takes that test img and makes it smaller than the original file, which results in faster download speeds. So the 1mb img for the speed test is actually compressed to smaller than 1mb.



Huh? The transfer data on dslreports is not compressable, period. Its downloading binary file data generated randomly , not an image that it can compress. This is a file transfer speed test, not timing an HTML page load-time which may be faster thru cingulars proxy.


Posted by: RF9

Quote:
Originally Posted by michaelflorio82
Just a question for you RF. In another thread I had posted pretty much what you said about img compression, and someone ekse posted this? Can you give me some insight to which one is correct?

Quote:
Originally Posted by michaelflorio82
Thats correct, except the proxy takes that test img and makes it smaller than the original file, which results in faster download speeds. So the 1mb img for the speed test is actually compressed to smaller than 1mb.



Huh? The transfer data on dslreports is not compressable, period. Its downloading binary file data generated randomly , not an image that it can compress. This is a file transfer speed test, not timing an HTML page load-time which may be faster thru cingulars proxy.
Well it appears I was wrong. It does not download an image that is compressed by the Proxy.
It's timing how long it takes to download a web page. That web page is made a certain size by embeding binary data of XXX KB inside that HTML page.
So there appears to be nothing compressable or that can be size manipulated to throw off the test like I have been erronously saying.

At some time in the past it did used to use an image like I had been saying, which was the last time I looked in to how MSpeed was doing the test. But it no longer does and I don't think it has for quite a while.

This doesn't mean the Proxy isn't affecting the test. It certainly still is. About the only think I know of that it can do is buffer the page download and speed up transactions between the handheld and the server.
Buffering means that the handheld doesn't need to go back to the web server for the next download segment, the proxy serves that (saving round trip latency.) And speeding up the transactions can shave a little latency off.

If anything it's taking the natural latency out of the mobile network and optimizing the test. But it appears the test is still valid since you're still downloading all of the content in a given period of time.

I guess I'm wrong about the proxy increasing latency. If anything it's really optimizing page download and is probably a benefit to people more than anything.

I suspect in the last year they upgraded the proxy. It used to really suck and really slow things down. But with the launch of BJ, 750, and now the new 8525 ROM which all use the proxy. They beefed this up.

I guess you can trust even proxy numbers. But I still think you need to turn off the proxy to get a true test of your bandwidth performance. I really don't know definitively what the proxy is or could be doing to the test.

I hope I cleared things up.... as clear as mud anyhow.

I used to work on the Netscape Proxy years ago (around 98,0 which is where I get these theories from.




vBulletin Copyright ©2000 - 2009, Jelsoft Enterprises Limited.
vB Easy Archive Final ©2000 - 2009 - Created by Stefan "Xenon" Kaeser