The Next Internet
So we had a guy come to Google to talk about a new way to think about networking. The existing implementation and structure of the Internet is sufficient, but not well-suited to most of the use of the internet today. The problem is that the Internet is all about connecting point A to point B. Then they talk, share data, whatever. But most of the time you just have a few publishers (web sites) sharing the same data with thousands and millions of users. This results in a lot of redundant bits being thrown all across the Internet, sometimes routers can be servicing thousands of users, and all of them are getting the same data. That's a huge waste of potential bandwidth.
He drew a clever analogy with the original internet. Originally, the internet was built on top of the phone lines. The phone network was all about creating a circuit. It wasn't about a phone-call, it was about constructing this circuit across large backbones. Phone numbers weren't addresses, they were the programming to connect one wire to another. When the original networking people started sending computer bits over the phone wires, everyone was up in arms! This wasn't how communication is to be done! Everyone knew that communication was about circuit building. It was ridiculous to build this ad-hoc, failure-prone, best-effort network. That couldn't go anywhere!
Similarly, the new revolution today should be about data. People should just be able to specify what data the want, and it shows up. They don't care about where it comes from. The only reason to care about data's origin is because that's how the security model is based. No, I don't "trust Microsoft", but I do want to know that this is the update which Microsoft has published, and not some hacked version. Interestingly the best work going on in this field is peer-to-peer services. Bit Torrent is pretty good, but still not good enough. We shouldn't need to depend on trackers.
The internet is about data now. That doesn't make the two-way communication TCP/IP layer it was built upon obselete, it just shouldn't be the focus of research or advancement. It's done it's duty, now we shouldn't care. Just like we don't care what on top of which physical medium IP is overlaid.
You can watch the video here.
And now, for another use of the Internet, IM
He drew a clever analogy with the original internet. Originally, the internet was built on top of the phone lines. The phone network was all about creating a circuit. It wasn't about a phone-call, it was about constructing this circuit across large backbones. Phone numbers weren't addresses, they were the programming to connect one wire to another. When the original networking people started sending computer bits over the phone wires, everyone was up in arms! This wasn't how communication is to be done! Everyone knew that communication was about circuit building. It was ridiculous to build this ad-hoc, failure-prone, best-effort network. That couldn't go anywhere!
Similarly, the new revolution today should be about data. People should just be able to specify what data the want, and it shows up. They don't care about where it comes from. The only reason to care about data's origin is because that's how the security model is based. No, I don't "trust Microsoft", but I do want to know that this is the update which Microsoft has published, and not some hacked version. Interestingly the best work going on in this field is peer-to-peer services. Bit Torrent is pretty good, but still not good enough. We shouldn't need to depend on trackers.
The internet is about data now. That doesn't make the two-way communication TCP/IP layer it was built upon obselete, it just shouldn't be the focus of research or advancement. It's done it's duty, now we shouldn't care. Just like we don't care what on top of which physical medium IP is overlaid.
You can watch the video here.
And now, for another use of the Internet, IM
posted by Alden | 22:15
0 Comments:
Post a Comment
<< Home