How is that insightful and not flamebait? I very much enjoy Firefly totally independantly of the first 3 seasons of Buffy (and despite the last season).
C64 scene wants your 360k disks
on
Homeless Wires?
·
· Score: 1
If the purpose is to speed up web access, then why couldn't all this gzip compression, prefetching, and so forth, be handled on your local drive without going through Google? Wouldn't that be faster?
Not necessarily. Here's an example;
Let's call the page you want to view the "source". We'll assume that the connection between the source and Google's servers is faster than the connection between the source and your PC. Only for extremely local (to you) pages is this likely to not be the case.
Let's say the source can provide a gzipped version of the page you want, with nicely optimised JPEGs. This is the best you can hope if you simply adhere to web standards.
Now let's say that Google's servers and Google's web accelerator can communicate using a compression system better than gzip -- there are plenty -- including a system that can pack JPEGs (arithmetic encoding saves around 12%, Stuffit v9 saves around 20%).
Assuming that Google's compression system save a further 15% over whatever the best web standards are, you will see a significant saving even if the packets are taking a more scenic route.
You gain even more saving with differential downloading. What's more, if you lose the stupid prefetch junk, this system even saves time on dial-up accounts -- depending on just how smart the compression and diffing is.
Re:Hidden tidbit in your post
on
Gates on Google
·
· Score: 1
I like Windows. The core of Windows. The basic platform for device drivers and UI. Not a problem with it at all. I can make it dance. It's all the other junk that's bundled with it I rip out. And I only use Office because at work I have to.
Re:Hidden tidbit in your post
on
Gates on Google
·
· Score: 1
If I were Microsoft, I'd drop all my unprofitable departments and focus on what people actually want.
Yes, but you'll need another proxy that accepts non-localhost connections. I recommend Proxomitron. Then set Proxomitron to accept connections from your network, and forward them to the webaccelerator.
This is a good idea on any home network with multiple PCs, Linux, Windows, Mac or whatever. This puts a single high-tech cache your side of your Internet connection talking to a single, massive, high-tech, compressing, optimising cache on the other side of your network. I think it's time for my home network to gain an always-on, fanless, Windows PC.
I can't imagine they've improved much with the caching, but if it's true differential downloading, your web surfing will get much faster. As for compression, there are far more efficient things in the world than mod_gzip, or for that matter JPEGs. The people that now own Stuffit (Allume) have got a nice program that can compress JPEGs by roughly 20-25% -- one hopes they're scrambling to offer something to Google.
This is a caching proxy with some pre-fetch stuff bundled with the plug-in. It will consume more bandwidth at the beginning, but as it caches your regular pages you should notice the traffic drop. Hopefully there's some nice compression between the local proxy and Google's server that will make the initial impact a little lighter, and the long term gains more siginificant.
Browser -> Localhost:8080 (Proximitron with ad filtering) -> 192.168.3.3:8080 (Proxomitron acting as a relay to jump networks) -> 192.168.1.1:3128 (Squid) -> The Internet.
When I did a default install on my local PC, the following was formed;
Browser -> Localhost:8080 -> Localhost:9100 -> Google's server -> The Internet
Which works, but now has traffic going through the "wrong" ADSL connection. Then I worked out that I could setup Google's webaccelerator further down the chain;
Browser -> Localhost:8080 -> 192.168.3.3:8080 -> 192.168.3.3:9100 -> Google's server -> The Internet
I'm not positive that 3.3 is routing through the 192.168.1.1 gateway, but if I really cared I'm sure I could setup a default route for Google's IP range. What I'm more likely to do is suggest that staff start using the 192.168.3.3:8080 proxy, which will automatically route them through Google's accelerator. After I explain the privacy implications first, of course.
The partnership between Google's server and Google's local software can produce a better optimised stream than anything you can do purely locally. Basically, Google copes with that initial sloppy junk from the source website, after that it's all streamlined using whatever modern compression techniques Google feels like using. No waiting around for new standards to be formed, they can implement more efficient compression systems as soon as they're available.
True Internet accelleration is easily achievable with a browser plug-in/server pair. Other than that, your best bet is a good ad filter. What's more, they stack.
I just realised that rather than installing it on the local PC, you can install it on any Windows PC with an existing proxy, and redirect all your requests from anything through it.
I'm watching my local proxy's logs and I'm seeing a request every time I mouse over a link. Can anyone confirm this behaviour? Sounds like it would slow things down, not speed them up.
Google already caches a huge amount of the 'Net. Why do you think this is so far away from their core business? I'd say that Gmail was more of a stretch than this.
Well, I've installed it on this and my own personal ad filter is still working. Requests appear to be directed at localhost:9100, but still through the proxy configured in Firefox's settings. I have had to bypass a remote proxy I used to use. This has a few negative implications and I'm not sure if I'll see a net reduction in traffic through our ADSL connections. But I'm a compression junkie, so I certainly want to give it a go.
I subscribe to Opera's compression/optimistation proxy for their mobile web browsers (I have Opera on my N-gage). The concept is simple and sound. A browser and proxy pair can adopt techniques that are not general standards. The boring, original page is fetched by the proxy, then highly optimised for the (known/modified/extented) browser. The result is a smaller download.
To give an example, Google could implement Allume's JPEG compression on images such that the JPEG data your browser now downloads from Google's server is 20-25% smaller. No standard proxy or browser sould support this, but Google's proxy and Google's Firefox extension could.
Well, I use the proxy product "Proxomitron" and it has the option to either only listen to the local PC (default), or open it up to an IP range (I use this on a server to bounce requests through a separate network -- or I did, the Google Web Accelerator product doesn't work if I do this, and appears to have no way to get it to pass its requests through a proxy.)
OSS has no greater chance of being the target of a patent, or copyright, lawsuit than closed source software. In fact, any organisation can view the source and make their own risk assessment; something you can't do with closed source software.
Slashdot Mirror
At least this is new. It's been a while since the last new insane bit of grammar/spelling I've seen ("copy'd").
Even the account creation system is down: "133: Internal Resource Exception."
How is that insightful and not flamebait? I very much enjoy Firefly totally independantly of the first 3 seasons of Buffy (and despite the last season).
Try Lemon64.com, look for the forums.
- Let's call the page you want to view the "source". We'll assume that the connection between the source and Google's servers is faster than the connection between the source and your PC. Only for extremely local (to you) pages is this likely to not be the case.
- Let's say the source can provide a gzipped version of the page you want, with nicely optimised JPEGs. This is the best you can hope if you simply adhere to web standards.
- Now let's say that Google's servers and Google's web accelerator can communicate using a compression system better than gzip -- there are plenty -- including a system that can pack JPEGs (arithmetic encoding saves around 12%, Stuffit v9 saves around 20%).
- Assuming that Google's compression system save a further 15% over whatever the best web standards are, you will see a significant saving even if the packets are taking a more scenic route.
You gain even more saving with differential downloading. What's more, if you lose the stupid prefetch junk, this system even saves time on dial-up accounts -- depending on just how smart the compression and diffing is.Didn't Microsoft just *lose* a patent lawsuit?
I like Windows. The core of Windows. The basic platform for device drivers and UI. Not a problem with it at all. I can make it dance. It's all the other junk that's bundled with it I rip out. And I only use Office because at work I have to.
If I were Microsoft, I'd drop all my unprofitable departments and focus on what people actually want.
This is a good idea on any home network with multiple PCs, Linux, Windows, Mac or whatever. This puts a single high-tech cache your side of your Internet connection talking to a single, massive, high-tech, compressing, optimising cache on the other side of your network. I think it's time for my home network to gain an always-on, fanless, Windows PC.
- Compression
- Caching
- Differential downloading
I can't imagine they've improved much with the caching, but if it's true differential downloading, your web surfing will get much faster. As for compression, there are far more efficient things in the world than mod_gzip, or for that matter JPEGs. The people that now own Stuffit (Allume) have got a nice program that can compress JPEGs by roughly 20-25% -- one hopes they're scrambling to offer something to Google.This is a caching proxy with some pre-fetch stuff bundled with the plug-in. It will consume more bandwidth at the beginning, but as it caches your regular pages you should notice the traffic drop. Hopefully there's some nice compression between the local proxy and Google's server that will make the initial impact a little lighter, and the long term gains more siginificant.
Browser -> Localhost:8080 (Proximitron with ad filtering) -> 192.168.3.3:8080 (Proxomitron acting as a relay to jump networks) -> 192.168.1.1:3128 (Squid) -> The Internet.
When I did a default install on my local PC, the following was formed;
Browser -> Localhost:8080 -> 192.168.3.3:8080 -> 192.168.1.1:3128 -> Localhost:9100 ! Fail !
So I turned off the jump to 3.3;
Browser -> Localhost:8080 -> Localhost:9100 -> Google's server -> The Internet
Which works, but now has traffic going through the "wrong" ADSL connection. Then I worked out that I could setup Google's webaccelerator further down the chain;
Browser -> Localhost:8080 -> 192.168.3.3:8080 -> 192.168.3.3:9100 -> Google's server -> The Internet
I'm not positive that 3.3 is routing through the 192.168.1.1 gateway, but if I really cared I'm sure I could setup a default route for Google's IP range. What I'm more likely to do is suggest that staff start using the 192.168.3.3:8080 proxy, which will automatically route them through Google's accelerator. After I explain the privacy implications first, of course.
The partnership between Google's server and Google's local software can produce a better optimised stream than anything you can do purely locally. Basically, Google copes with that initial sloppy junk from the source website, after that it's all streamlined using whatever modern compression techniques Google feels like using. No waiting around for new standards to be formed, they can implement more efficient compression systems as soon as they're available.
I imagine there's a hash exchange between your local cache and the server to determine what version you've currently got.
True Internet accelleration is easily achievable with a browser plug-in/server pair. Other than that, your best bet is a good ad filter. What's more, they stack.
/heads off to install it on one of the servers.
I'm watching my local proxy's logs and I'm seeing a request every time I mouse over a link. Can anyone confirm this behaviour? Sounds like it would slow things down, not speed them up.
Google already caches a huge amount of the 'Net. Why do you think this is so far away from their core business? I'd say that Gmail was more of a stretch than this.
Well, I've installed it on this and my own personal ad filter is still working. Requests appear to be directed at localhost:9100, but still through the proxy configured in Firefox's settings. I have had to bypass a remote proxy I used to use. This has a few negative implications and I'm not sure if I'll see a net reduction in traffic through our ADSL connections. But I'm a compression junkie, so I certainly want to give it a go.
To give an example, Google could implement Allume's JPEG compression on images such that the JPEG data your browser now downloads from Google's server is 20-25% smaller. No standard proxy or browser sould support this, but Google's proxy and Google's Firefox extension could.
Well, I use the proxy product "Proxomitron" and it has the option to either only listen to the local PC (default), or open it up to an IP range (I use this on a server to bounce requests through a separate network -- or I did, the Google Web Accelerator product doesn't work if I do this, and appears to have no way to get it to pass its requests through a proxy.)
OSS has no greater chance of being the target of a patent, or copyright, lawsuit than closed source software. In fact, any organisation can view the source and make their own risk assessment; something you can't do with closed source software.
Is there an ebook version?