Slashdot Mirror
Intel says Internet needs to change
Nurgled writes "At a recent Intel conference, CTO Pat Gelsinger said that something needs to be done to avoid the Internet buckling under the strain of new technologies and millions of new users. The BBC reports that Intel is attempting to layer a 'new Internet' over the existing network which can detect and counteract things like worm outbreaks and route traffic more intelligently during low and high traffic periods. Intel's prototype, PlanetLab, has 441 nodes but claims to be an open platform with documentation available on the site. What's in it for Intel, though?"
The internet could use some change, I just got another "Nothing to see here, please move along." message.
I don't like the idea of Intel owning the internet
Why not just boost adoption of IPv6?
I dunno; maybe they like using the internet? Intel may be an Evil CorporationTM, but they've got as much interest as anyone else in keeping it going.
Or maybe - just maybe - they're doing something nice.
Then again: to quote the article
" If the net grows to 100 billion devices connected to it, our goal is to have a piece of Intel inside in every one of those hundred billion "
Pat Gelsinger, Intel
The problem is, a lot of the internet is dated from way back. You not only see it in 'the internet' and the main protocol being used (tcp/ip, which is, as far as I can see, the thing intel wants to change), but also how some applications talk to each other.
For example, the SMTP protocol. It was designed WAY back, and only a few people had problems with not being able to verify the sender of an email, but that was being ignored. If someone would want to make such a protocol nowadays, it would contain a HELL lot more security measures. But if you want to change the protocol right now, you will need a pretty big front of important people in order to do that...
My point is: Intel can say they want to make a new layer on top of the internet, which is all fine, but I think in order to really make a 'better' internet, you need to change the way application communicate with each other too...
- Leon Mergen
http://www.solatis.com
So Intel will reinvent Internet2 then?
Why don't they just support and push the adoption of IPv6 and build it correctly from the ground up vs. changing whats already in place?
It's called SkyNet.
---
IMHO, of course.
May the SOURCE be with you.
http://stephan.sugarmotor.org
Obvious is the answer: total domination of the next generation of technologies. Intel realizes that microprocessors, the market on which it built its business, is fast becoming a mature industry. Margins will drop as competition between AMD64 and Intel64 heat up. In search of new areas of grow, Intel is branching out into other areas: routers, WI-FI, etc.
Intel does nothing out of generosity. More than 30% of the company is H-1B workers, and they retain the same ruthlessly competitive attitude that they had in their homelands (e.g. China, India, etc.)
What's in it for Intel is to sell chips to power said 'new internet'. How dare they.
I like Intel's new concept. Alot of the internet is still running old hardware, you can't just change the internet, because it's impossible to change all the routers/etc at once but if they do it this way, they can just wait for the old internet to deteriorate.
but does this article have anything to do with "Internet2"? i'm a little confused, because the description sounds like Internet2.
Let us know when your organization is part, then we can join your organization instead of going to the consortium directly.
Or maybe slashdot.org can sign up.
Stephan
http://stephan.sugarmotor.org
... built a GUI on DOS and called it Windows and that sucked. What the internet needs right now is more and more bandwidth both on the backbone networks, in the near future more bandwidth to individual workstations and maybe in five years from now IPV6. What the Internet does not need is censorship, TCPA/Palladium Digital Rights Management, Taxes, Microsoft and least of all Intel.
The entire PlanetLab system runs on-top of Linux, I would say that is most certanily an open platform. All their tools include source code, it quite possible to make PlanetLab run on other platforms also.
Intel does have an interest in keeping the Internet going, they just might be foreward thinking enough to realize that if they want to stay in business they are going to have to support things like this even if it does not have immediate financial returns.
Also, Mozilla has to run on somebody's chips, Intel is not a software vendor, all that matters is if people keep buying their chips. Also, Intel will pour money into this because if it is the next big thing they suddenly understand it the best of anybody, and then you have a solutions provider business model, one that works very well.
Intel, by providing free and open standards will showcase themselves as a pioneer willing to make sacrifices to maintain the leadership role their company currently has. Nothing lasts forever and if they think only with greed they will more easily lose their "number one" status.
In general, you want to keep the field you play on in good shape. You need to take care of your arena so people find value in your products. If the Intel research will make internet use greater for more people, this directly benefits Intel as it will lead to presumably more chip sales in the end.
If they really get something good going here and fail to keep it open and free, no one will adopt it and they will have just wasted money on research that will not pay off and not have increased chip sales.
Then again, I could be entirely wrong here and Intel needs to figure out a way to increase their already huge profit margins. This may be the way?
"If you are a dreamer, a wisher, a liar, A hope-er, a pray-er, a magic bean buyer
Maybe if they had a track record of creating open standards I could believe that they wouldn't corrupt it, but I don't think this would be a good idea. I doubt it will catch on.
Intel seems to think that networks need to get smarter. But networks need to get dumber (i.e. more simple). Systems need to be more like OpenBSD and less like [bloated] Linux or Windows. Applications need to be smaller and more precise.
As everything becomes more and more embedded, we need to strip functionality that we don't use anymore and build applications to what we do, not what we did five (or ten, or twenty) years ago.
Open-source has always strived to provide less bloated and overall better quality software. This comes from the Unix mentality. Intel does not yet understand this approach to computing. Intel provided a hardware architecture that rivaled IBM for monolithic and for lack of innovation and growth. This is mostly thanks to Microsoft and the users of Microsoft products.
We in the open-source world and of the Unix generation have never had severe problems with viruses. We learned from the mistakes of the [original] Internet worm, and we haven't made those same mistakes again. We don't neet smart networks. We need streamlined networks, systems, and applications. Small progams with single purposes: to do one thing well.
IPv4 allocations for hobbyists? join the ipalloc-l mailing-list! www.operations.net/mailman/listinfo/ipalloc-l
This has nothing to do with "scaling". It has everything to do with re-inventing the technology so that they control. Basically, Intel is not able to take on Cisco directly, so instead will attempt to shift the playing field to their backyard.
I prefer the "u" in honour as it seems to be missing these days.
What do you expect? Their stock is stuck in a rut and their products have become commoditized and China Inc., begins to play on their court.
They need something that looks like a new huge market to try to stem the bleeding and loss in investor confidence.
Its unlikely that the likes of Cisco and Juniper and Huwei are simply going to stand still anytime soon. Indeed Cisco just indicated that they will attempt to double their product offerings and rate of introduction of new products over the next 5 years. Juniper continues to move forward on the high end and Huwei is busy outpricing everyone worldwide on the low end and begining to ramp up into higher end products. The PC market is stalled as our president has successfully diverted much IT spending toward paying for higher energy and borrowing costs.
Current investments in existing infrastructure including the steamroller of lobbyists behind the new internet 2 roll out are out of Intel's control. The core of their business model is now under attack by AMD and its Opteron, so announcements like this are critical for them to keep their heads above water.
The real issue here is whether they can win any of the super-scret contracts to route and anlyze all internet traffic through the new NSA mainframe filters that are straining to keep up with the explosion of foreign and domestic internet use or whether they can win any of the contracts larger corporations are now issuing to keep track of everyones internet device and VOIP use on a 24/7 basis. Now that is where the real new growth in the market is not on selling to the few folks who still have a little money to spend on IT.
The article is very misleading. PlanetLab is primarily developed by Princeton University. The claim that the network was "funded by Intel" is a huge exaggeration: Intel has merely donated some of the original servers, which are now only a fraction of the total number of PlanetLab servers.
t lab.htm
You can read a more informative article about the background here:
http://www.princeton.edu/pr/news/03/q2/0624-plane
Furthermore, I'd like to point out that most of the work on the PlanetLab infrastructure is done by grad students at Princeton University, not by Intel.
If we are talking about mandatory authentication, then there needs to be some way to securely authenticate. We have optional authentication now, which is good, but too easy to circumvent. Secure authentication requires a protocol and secure hardware and software. Both are right up the Wintel alley, with thier embedded ID chips and closed OS. Again, the protocol could be open and free, but only certified machines are allowed ont he network. Will certification be anything other than a $50 bill slipped to Intel. Maybe not.
"She's a scientist and a lesbian. She's not going to let it slide." Orphan Black
I think the frame holding your MCSE cert is crooked.
'Buckling under the strain of new technologies'? What about buckling under software patents?
I see the major flaw in Intel's idea right there where they say their network would be optimized for web services. Umm, what if web services aren't the big thing in a couple of years? IBM optimized SNA for the kinds of networks they knew were going to be built, but we don't see much SNA outside mainframe data centers. Corporate America doesn't have a very good track record of predicting how things will actually turn out. The Internet's strength is that it isn't optimized or designed for any one application, so while it may not be ideal for any one application it's at least usable for all of them. I'd be wary of changing that.
As far as worm outbreaks, those don't require fundamental changes in the Internet. Stopping them requires the people responsible, the ordinary users, to get a clue. It doesn't even have to be much of a clue, just something on the level of "Running that red light at a busy intersection might be a bad idea." or "Putting my hand on the red-hot stove burner might hurt.". That's not asking that much. (NB: that wasn't in the nature of a question.)
Of course it is Intel that needs to change. Hardly anyone cares about PC cpu MHz anylonger. The Itanium is such a magnificent failure. PC and server CPUs will soon cost less than $50, and nobody will care about brands anymore.
What should Intel do? They have to do something that makes the market believe Intel is at least part of the future. Pushing that Internet needs to change seems to be a way to get heard at all.
Maybe Intel is part of the future - and maybe they will revolutionise Internet. But primarily it Intel that needs to change - not the Internet.
It's a bit off-topic, but if you want to help quash viruses and the like, incorporate a basic ip4/ip6 firewall in NIC chipsets.
n -source BIOS projects is welcome to take these ideas and run with them. Granted, if it's in the system BIOS rather than the NIC BIOS it may only work with 1 NIC and as such, be only a proof of concept.
Home-user PC manufacturers could set to to
"block all unrequested inbound traffic and block all outbound traffic except:
web, ftp, ssh, dns, bootp, tftp, dhcp" and maybe a few others, and provide a web interface where the 1st question after "please enter password" is "who is your email provider" to open up email ONLY to that location. Better yet, if the email provider isn't configured, beep during POST and give the user an opportunity to enter the NIC-bios-setup screen to set it.
Of course, it would need to be at least as configurable as the firewalls built into most "home routers."
The technology to do this is already there, and you can argue it's already been done given that a PC with a network-interface and a software firewall amounts to the same thing, and it's "obvious" that such a system can be burned to firmware. As such, any patents would be narrow and probably serve only to prevent cloning of a specific chipset.
Anyone working on any of the http://www.aloha.com/~knowtree/links.html#BIOSope
Knowledge is how to play a game, intelligence is how to win, wisdom is knowing what game to play.
"Hey, I know how to fix the internet!"
"How?"
"Build a new internet on top of the old one!"
"Uhh..."
The internet doesn't need fixing, it seems to run just fine. What it does need are less people running virus magnets and creating all kinds of problems. The lack of security is *not* the internet's fault; it already does what it needs to do. Security should not be the job of the transport. The job of the transport is to transport stuff, be it unencrypted data or the next generation of uber-encrypted VPN for those who want security. (This is my gripe with all these "wireless security" methods. Just build a damn base station with a built in VPN server and be done with it. But then they couldn't introduce "new and improved security" every other month and sell more stuff.)
Got virus problems? It's not the internet's falut, nor its responsibility. The responsibility for that should be on the client side. I see attempted windows exploits coming to my network all the time: in my denied connections for my firewall. Packets dropped, no harm done. Same with my Apache logs. I scan my incoming and outgoing email for viruses, firewall everything, and make those in my family who use Windows aware of issues like don't click on random shit in your email you know nothing about. And guess what? Everything works smoothly and plays nice.
The idea is nothing more than buzz to create some interest from people who have money in the hope that they'll part with their money. "Look what we can do, we can fix the internet! Now, we'll just need you to write a check for..." Besides, how long will it be before an internet tunneled over an internet gets overloaded? Then what? Tunnel another internet over the internet tunneled over the internet? If you want a new, faster, better network, you gotta build one from the ground up.
this is my sig
Spam is a social problem, and it has to be solved through social mechanisms. Every time a technological fix to spam has been developed, the spammers have found a way to get around it.
It doesn't matter that the spammers are transnational, the biggest spammers are all in the US, no matter where their servers are, because it's a rich country with weak privacy laws. For the forseeable future, big spammers are going to want to live in rich countries, and they're going to want to operate from countries where their very databases aren't illegal. If there were US laws that addressed the behaviour that causes the problem without loopholes for 'well behaved' spammers, and these laws were enforced, this WOULD reduce spam from a universal pollution to an annoyance.
This means: ban unsolicited broadcast email. This means: don't force people to opt out, don't make exceptions for popular spammers (we don't make exceptions that allow charities or political parties to hold regular "tire bonfires" in their parking lots), don't allow "properly labelled" spam, just ban UBE, commercial or not.
That means, if you're mailing to more than a few people (let's say, 100 copies of a message a week as a limit... that's plenty high enough for any legitimate purpose and far far below what a spammer needs to stay in business) then you better have (a) a verifiable signup record for each person if it's a mailing list, (b) record of an explicit request from each person, if it's a response mailing, or (c) a proof that you have an existing business, professional, or membership relationship that each recipient is in a position to terminate.
No exceptions.
PlanetLab is not "Intel's prototype". Intel did not start the project,
and has never been in control of it. PlanetLab is primarily an
academic project that receives funding from a number of corporations,
including HP, Google, AT&T, France Telecom, and Intel.
The steering committee consists of faculty members from four
universities along with one representative from HP and one from Intel.
The research staff is composed mainly of people from Princeton along
with at least one from Berkeley.