By offering to implement another application layer protocol Google is promising up to 100% improvement in internet page loads. This is totally independent of the current Chrome browser, in case you’re wondering.
That’s what is being claimed in a blog post by someone on the Google campus.
A story in ComputerWorld gives the changes that would be needed –
Google is hoping to make Web pages download up to twice as quickly using SPDY, a new application-layer protocol it’s experimenting with, the company said in a blog post.
It wants to improve on the performance of using HTTP (Hypertext Transfer Protocol) by minimizing latency. For the protocol to work, the browser and the web server have to be upgraded, but changes to Web pages are not needed, according to Google.
Google’s lab tests of SPDY show an improvement in page load times compared to HTTP of between 27% and 60%, and between 39% and 55% when using SSL (Secure Sockets Layer), it said.
The company still needs to do a lot of work to evaluate the performance of SPDY in real-world conditions, the blog post said.
The next part of the story talks about prioritizing requests, sending packets simultaneously, and a compression scheme, which immediately makes me wonder if we are not talking about a modified (or perhaps only modified enough to escape patent problems) Opera Turbo scheme.
Google conducted the tests by downloading 25 of the “top 100” web sites ten times each over simulated home network connections, using a prototype Google Chrome browser and a Web server that it has developed.
SPDY uses a number of techniques to speed Web downloads, including allowing many concurrent HTTP requests across a single TCP session, prioritizing the requests, and using compression to reduce the number of packets and overall amount of data sent.
Google doesn’t want to start from scratch with SPDY. The protocol still uses HTTP headers, but it overrides other parts of the protocol, such as connection management and data transfer formats.
“That Google is trying to improve download speeds is great, and the numbers are very promising,” said Måns Jonasson, a web developer at IIS, which is responsible for the top-level Swedish Internet domain, .se.
For something like SPDY to work everyone has to be on board. The protocol won’t become a success unless it’s supported in both Internet Explorer and Firefox, according to Jonasson. It might be able to convince Mozilla to implement the protocol in Firefox, but convincing Microsoft will be difficult, he said.
“Microsoft seems to hate everything Google does,” said Jonasson.
The source code for the prototype Google Chrome browser is available for download. The code for the server will be released as open source in the near future, it said.
We have seen Microsoft embrace open source lately (not that it has not always, as the very first Microsoft TCP/IP stack was ‘derived’ from the then current BSD stack), so perhaps that would not be such a problem. On the other hand, so what if Microsoft doesn’t adopt the changes? As its share of the browser landscape is shrinking all the time, it might be aware enough at this point to know that it must be less intransigent, in order to survive.
That Google is opening up the code is good, but it also needs to approach standards organization IETF (Internet Engineering Task Force), Jonasson said.
Development efforts on other protocols to speed Web downloads, such as SCTP (Stream Control Transmission Protocol) and SST (Structured Stream Transport), have seen little activity in recent years.
Perhaps it is time to take up the cause, and work on these abandoned projects, as they might yield better results, or perhaps be folded together, as something that provides a synergy for the final product.
When things like this occur, we are all reminded of the times when CPU manufacturers have run up against speed limits by merely raising clock speeds. When that happens, the engineers start looking for economy of overhead, and the ability to increase the number of instructions executed per clock cycle, along with larger amounts of data processed concurrently. By implementing this sort of aggregated method, perhaps we can see some very real gains in speed of current connections.
No one will be upset about that.
Quote of the day:
Never interrupt your enemy when he is making a mistake.
– Napoleon Bonaparte