back to article DEATH TO TCP/IP cry Cisco, Intel, US gov and boffins galore

The US National Science Foundation, Cisco, Verisign, Panasonic and boffins from around the world have thrown their weight behind a new “Named Data Networking Consortium” that aims to develop “a practically deployable set of protocols replacing TCP/IP that increases network trustworthiness and security, addresses the growing …

  1. Anonymous Coward
    Anonymous Coward

    Verifying the source of the data

    Very useful for those bodies who to want to track and intercept the data.

    Very useful for those bodies who want to claim licensing royalties for the use of such a protocol.

    1. Ole Juul

      Re: Verifying the source of the data

      Very useful for those bodies who want to claim licensing royalties for the use of such a protocol.

      Indeed, the article says "open source" but does not mention if it will be free. I'm rooting for a BSD license like TCP since that seemed to work very well.

    2. Nigel 11

      Re: Verifying the source of the data

      Track, yes, if "anonymous" is not built in as an acceptable source. But how is that different to today's internet, or the telephone network? If a two-way communication is desired, then an outgoing packet has to have a truthful source. If verification was built in, you'd be able to know that the claimed source was untruthful, and 99% of the time the right action to take for a non-verifiable source will be to drop the packet.

      Intercept, no. There is no conceivable way to prevent the data content of the packets being encrypted. By the way, one of the dual purposes of public-key cyphers is sender-verification. You can even do verifiable plaintext. (Transmit single-encrypted with your own private key, rather than double-encrypted with your own private key and your recipient's public key)

    3. NoneSuch Silver badge

      Re: Verifying the source of the data

      American developed "standards" are what got us here in the first place. Anything that is approved by the US Gov should be shunned by the rest of the world. We need to look at the stuff the yanks disapprove of.

      1. Anonymous Coward
        WTF?

        Re: Verifying the source of the data

        American developed "standards" are what got us here in the first place.

        Eh?

        Korea's Anyang University, China's Tongji and Tsinghua Universities, the University of Basel (Switzerland) and Japan's Waseda University are also aboard the effort.

        Huwawei are Chinese

        Alcatel-lucent are French.

      2. Jamie Jones Silver badge

        Re: Verifying the source of the data

        "American developed "standards" are what got us here in the first place. Anything that is approved by the US Gov should be shunned by the rest of the world. We need to look at the stuff the yanks disapprove of."

        Huh? Are you saying that TCP/IP isn't a success story? From its humble beginnings to where it is now, used on a scale unimaginable at it's inception, designed for a totally different user environment?

        1. Trevor_Pott Gold badge

          Re: Verifying the source of the data

          "Huh? Are you saying that TCP/IP isn't a success story? From its humble beginnings to where it is now, used on a scale unimaginable at it's inception, designed for a totally different user environment?"

          There are far more important things in this world than commercial success. Privacy, anonymity and civil liberties are great examples. Both are things that TCP/IP has fantastically failed to deliver, and IPv6 has completely eliminated.

          TCP/IP's time is past. It is time now for something designed from the ground up to ensure privacy and anonymity as a means of helping internet citizens retain their civil liberties.

          1. Jamie Jones Silver badge

            Re: Verifying the source of the data

            Trevor, I totally agree with you about privacy/anonymity/civil liberties.

            I was just pointing out that whichever way you cut it, TCP/IP has been a success story (and I didn't say 'commercial' success, although it's probably implied somewhat, in that the internet companies wouldn't exist if there was nothing in it for them)

            Cheers, Jamie

            1. Trevor_Pott Gold badge

              Re: Verifying the source of the data

              See, I can't agree. It has been a success by some measures. I simply disagree that "uptake" is the only relevant measure.

              The fact that they can't get privacy right - and in fact nerfed the shit out of it in v6 - says to me it is as much of a spectacular failure as it is a success. The two can - and do - exist in tandem.

              1. Jamie Jones Silver badge
                Happy

                Re: Verifying the source of the data

                @Trevor, fair enough, I concede there are many ways it's been unsuccesful too, but I would still maintain it's an overall success story. (We've managed to disagree without insulting each others mothers..... that's an internet first!)

                1. Trevor_Pott Gold badge
                  Pint

                  Re: Verifying the source of the data

                  I think there is a middle ground here. I agree that TCP/IP was a success, however, I feel it's day has passed. The domestication of horses for use in hauling cargo was a success for much of human history. Eventually, however, we needed to haul more than could be hauled by horses. The train was invented. We have cargo boats and semi trucks and full bore size-of-bloody-houses mining trucks.

                  Given the mistakes and failures of TCP/IP I think it's absolutely time for something better. That doesn't detract from the past success of the protocol, but it does mean that it's time to stop trying to pull our mining equipment around with ever larger teams of horses.

                  Also: something something your mom. Because the internet. :)

              2. Charles 9

                Re: Verifying the source of the data

                "The fact that they can't get privacy right - and in fact nerfed the shit out of it in v6 - says to me it is as much of a spectacular failure as it is a success."

                No, it basically says the problem is intractable for two-way communications. The ONLY way to have any chance of privacy in communications is to make ALL communications one-way (basically, drop TCP for UDP) and make all listeners passive. This way, no one can tell (without additional clues) who's saying what to whom. Thing is, that's going to involve losses, overhead, inefficiencies, electricity, costs. Especially in situations where things like electricity can be at a premium. It leads to another intractable problem: that of creating communications that are simultaneous efficient and hard to fingerprint.

      3. Jaybus

        Re: Verifying the source of the data

        Eh? These are open standards published by IETF (International Engineering Task Force). The entire world has been "looking" at them for decades.

  2. Dave Harvey

    Anyone remember the OSI protocols?

    No - not just the conceptual "stack" - there were actually protocols designed in the early 1990s which some people expected to replace horrible, outdated TCP/IP.

    And we know how well they went down! In medical imaging, the standard was actually designed around them, so we have a horrible hack, emulating the OSI protocol.........over TCP!

    1. Anonymous Coward
      Anonymous Coward

      Re: Anyone remember the OSI protocols?

      The difference here is you have some massive heavyweights in the form of Cisco, Intel, Huawei, Acatel-Lucent and Qualcom in there. That is pretty much most of the networking market leaders right there.

      1. Warm Braw

        Re: Anyone remember the OSI protocols?

        Actually, most of the then heavyweights of networking (the PTTs) were behind OSI at the time. That was one of the problems - individual participants were able to insist, by virtue of their size, on the inclusion of features that suited their own interests, resulting in "standards" with so many options that interoperability was almost precluded by design.

        The more streamlined connectionless variant of OSI was heavily promoted by DEC (whose DECnet was at the time the system on which the world's largest networks operated) both in its product line and within standards bodies including ISO and the IETF.

        What essentially saved TCP/IP at the time was not the force of massive heavyweights, it was a chippy little upstart company, cisco, that could offer some immediate price-performance benefits assuming a 32-bit addressing scheme, and a severe case of "NIH" affecting the IETF.

        Never underestimate the power of short-term thinking.

        1. Dazed and Confused

          Re: What essentially saved TCP/IP?

          Was that it worked and it was good enough for what people wanted it for, whereas OSI gave infinite interoperability problems.

          Then came the PC, you could run TCP/IP in a system with 640K of base memory and still just about leave room for an app. This wasn't the case for OSI.

          Also I'm bloody glad I've not needed to wage war against X.400 in the last 16+ years.

          1. Mike Pellatt

            Re: What essentially saved TCP/IP?

            Also I'm bloody glad I've not needed to wage war against X.400 in the last 16+ years

            Looked at Exchange Server addresses lately ?? :-)

            1. Dazed and Confused

              Re: What essentially saved TCP/IP?

              > Looked at Exchange Server addresses lately ?? :-)

              Hey I run a Scalix server at home, which used to be OpenMail, so the whole internal addressing scheme is X.400. But I don't have to fight the rest of the damn thing and can largely ignore all the X.400 addressing stuff which was always bloody stupid and unfit for purpose.

          2. Gordon 11

            Re: What essentially saved TCP/IP?

            Was that it worked and it was good enough for what people wanted it for, whereas OSI gave infinite interoperability problems.

            Indeed. OSI was still developing theory - and being developed by committee....

            Also I'm bloody glad I've not needed to wage war against X.400 in the last 16+ years.

            Ah, yes. I once had to dissemble an X.400 message by hand to determine that HP would not let you pass a message between two MTAs in one organization unless they claimed to be from different organizations. HP themselves didn't seem to know they'd set it up this way.

            1. David Roberts

              Re: What essentially saved TCP/IP?

              What caused TCP/IP and SMTP to take over was the complete lack of security which made it trivial to implement and trivial to join and extend networks.

              With X.25 and X.400 you had point to point connections where both ends had to authenticate then agree what features they would use before you could get anything working. You had complex routing and charging arrangements because everything routed via PRMDs and ADMDs.

              TCP/IP and (optionally) a router plus the address of a DNS server and you could talk to anyone in the world almost instantly. Almost directly, as well - no going up and down hierarchies.

              As usual OSI looked off into the future and ignored the current day.

              Much of what was in the original OSI standards has crept in piecemeal via RFCs as the vision of the future proved to be correct (not that all of OSI was prophetic).

              If we did have a managed network where every messaging system had to be authenticated by and vouched for by an ISP then we might get rid of the massive SPAM and phishing email traffic.

              Field day for the bureaucrats and government spies, though.

              Also, the entry level cost would have kept home users off the networks.

          3. Warm Braw

            Re: What essentially saved TCP/IP?

            > it was good enough

            That's precisely my point. It was good enough at that point in time and the issue of what to do when it was no longer good enough was tossed into the long grass.

            That's the reality any "better" proposal has to face - see IPv6 (passim).

          4. Down not across

            Re: What essentially saved TCP/IP?

            Then came the PC, you could run TCP/IP in a system with 640K of base memory and still just about leave room for an app. This wasn't the case for OSI.

            Anyone who ever had to work with DEC Pathworks would attest to that.

            Ok admittedly it was mostly about LanManager, but it did support DECnet, LAT etc.

            Getting anything to run (even with QEMM) was a chore. Packet drivers and Trumpet was much nicer.

            1. Kubla Cant

              Re: What essentially saved TCP/IP?

              Pathworks!! Now there's a stupid name I haven't heard for years. You really haven't suffered until you've set up a load of diskless workstations booting off the network from a VMS server. Not only did the protocol stack have to be shoehorned into memory, the entire thing had to fit on a 1MB disk image.

              1. Down not across

                Re: What essentially saved TCP/IP?

                You really haven't suffered until you've set up a load of diskless workstations booting off the network from a VMS server. Not only did the protocol stack have to be shoehorned into memory, the entire thing had to fit on a 1MB disk image.

                Ah. My sincere apologies for resurrecting the memories that had been confined to the dark, nearly unreachable corner. I'll crawl under my rock and ...umm...dunno.. go boot a DECserver with MOP or something.

        2. Roland6 Silver badge

          Re: Anyone remember the OSI protocols?

          >What essentially saved TCP/IP at the time was not the force of massive heavyweights, it was a chippy little upstart company, cisco,

          No what made TCP/IP was that it was good enough, bundled for free with Unix on workstations and that code was available that enabled it to be ported to other kit eg. IBM mainframes and IBM PC's (the big challenge of the 80's wasn't Internet but simply getting all those PC's and workstations to share printers and terminal access to internal systems). And whilst all the major players had OSI stacks by the end of 1988, they had vested interests in their own networking suites, hence getting hold of them and deploying them wasn't made easy (a task also not made easy due to the number of OSI profiles that existed at the time). Cisco in the 80's was just one of several networking equipment vendors, hoping to make it to the big time.

          The problem NDN has is effectively the same as OSI, breaking into a established market. Unless it can work over the established IPv4 infrastructure (or possibly IPv6) it will rapidly become just an interesting research project.

      2. Mike Pellatt

        Re: Anyone remember the OSI protocols?

        Not all the names you list weren't interested. Intel were well in there on OSI. They had a guy in EMEA, based in Swindon, dedicated to MAP/TOP. DEC, too, as someone else pointed out. And ICL.

        From what I could tell later on when my then-company had an ICL connectivity product, was that it was the lack of mature routing and nameservice vis-a-vis TCP/IP that really did for OSI. Compared to TCP/IP, when properly configured, it ran sooooo much better over mobile networks.

        Of course, we could argue how mature TCP-IP's routing is when core routers need 512KB+ routing tables....

      3. Roland6 Silver badge

        Re: Anyone remember the OSI protocols?

        >The difference here is you have some massive heavyweights in the form of Cisco, Intel, Huawei, Acatel-Lucent and Qualcom in there. That is pretty much most of the networking market leaders right there.

        Not nearly enough support: Remember OSI in it's various guises had the backing of computing and network equipment vendors, Telcos, Governments, major end user organisations etc. and yet it still failed. As I've noted elsewhere, all the major vendors demonstrated 7 layers of OSI in 1988; practically within a year it had been forgotten about and everyone was buying TCP/IP. Even government procurement was buying TCP/IP as long as the proposal made suitable commitments to using Open Standards and OSI at some undefined future date.

    2. Phil O'Sophical Silver badge

      Re: Anyone remember the OSI protocols?

      we have a horrible hack, emulating the OSI protocol.........over TCP!

      Not that much of a hack, probably RFC 1006 "ISO Transport Service on top of the TCP", it dates from the late 80s (OSI is much older than 1990's)

      1. Dave Harvey

        Re: Anyone remember the OSI protocols?

        >>we have a horrible hack, emulating the OSI protocol.........over TCP!

        > Not that much of a hack, probably RFC 1006 "ISO Transport Service on top of the TCP", it dates from the late 80s (OSI is much older than 1990's)

        No - they decided to write their own from scratch - see DICOM Part 8.

    3. ecofeco Silver badge

      Re: Anyone remember the OSI protocols?

      Wow! Now there's something I haven't heard about in years!

      1. Anonymous Coward
        Anonymous Coward

        Re: Anyone remember the OSI protocols?

        > Wow! Now there's something I haven't heard about in years!

        It's like Fortran and COBOL, everyone loves to sneer at it as being so last century, but I think people would be surprised at just how much OSI networking there still is, especially in the finance and government world. We put our OSI products into maintenance a decade ago, and the sales folks still get calls from customers wanting licenses for newly-installed systems. Some of the support guys are kept busy making patches to keep them running on the latest OS versions (mind you, using the word "patch" to mean "a compete set of new 64-bit drivers" is maybe stretching it a bit...)

      2. Anonymous Coward
        Anonymous Coward

        Re: Anyone remember the OSI protocols?

        ISIS protocol is the IGP for many Service Provider networks. That's OSI based. It is used to distribute network reachability information (usually subnets for IP, but other for SDH) but it's not based on or using IP.

        X500 is widely used too. Not networking but it is open standard.

    4. yoganmahew

      Re: Anyone remember the OSI protocols?

      And then there's SNA... which required a separate MVS instance to run the handshaking and pathfinding. For us mainframers, having the comms stack on the box and communicating via an OSA card was a major improvement in comms speeds and device support/interoperability.

      Of course, lots of airline IP still pretends to be SLC... (MATIP).

  3. Bronek Kozicki

    interesting

    .... but I will remain sceptical until open source implementation appears and is merged into both BSD and Linux (and Windows just for laughs - we all remember NetBEUI and NWLink, right?)

    1. Anonymous Coward
      FAIL

      Re: interesting

      NetBEUI wasn't not MS so no idea why you are singling them out they just supported it. It was an IBM job.

      NWLink was a way of getting Windows to talk to Novells interpretation of IPX/SPX. The clue is in the name NetWareLINK. It actually did this pretty well. heck it was often easier to get MS kit talking to Novell kit than it was getting Novell talking to Novell.

      I guess you see the word Microsoft and the steam just starts billowing out of your ears.

      1. Anonymous Coward
        Happy

        Re: interesting

        Note to self. Leave thread for 2 minutes. Proof read.

  4. oldtaku Silver badge
    Unhappy

    And of course it'll have content DRM built right in. That's the primary reason they want this. HDCP-alike right at the network and transportation layers.

    1. Suricou Raven

      It's worse than that. If this system allowed anyone to utilise it to speed up their network, it'd become the greatest tool of piracy since the invention of usenet binaries. Anyone who could rent a £10/m VM could happily host a few terabytes of downloads at minimal cost - every dodgy 'free movies' site in Russia would be doing it. The proposed architecture does include a means of verifying the publisher of data, probably to handle exactly this situation: ISPs can make sure their new caching network system only accepts the content of 'reputable' publishers (Those with money or influence enough to get whitelisted) while everyone else has to make do with the old, slow, inefficient and expensive methods we use today.

      It also allows for geographic restrictions to be enforced by the network - eg, the BBC could tag iPlayer video content as 'UK exclusive' and routers for US ISPs would simply refuse requests for download. And Netflix could do the same to block their US-exclusive content over here. It'd be more reliable than geo-IP, which depends on a constantly changing and error-prone database.

      1. Anonymous Coward
        Anonymous Coward

        Except how would that stop proxies and crypto-wrapping, which are already used to get around IP detection?

        1. BitDr

          I would think that by its very nature (targeted entity) there are no proxies.. your device is the targeted "entity".

          1. Charles 9

            There's ALWAYS proxies. Given encryption, there's no way to block an encapsulated forwarding packet (the "double envelope" as I call it). As far as the ISP knows, the proxy is the targeted "entity", after which it's out of its control.

  5. Extra spicey vindaloo

    With incredible uptake of IPv6

    This consortium must really enjoy pissing into the wind!

    1. Michael H.F. Wilkinson Silver badge

      Re: With incredible uptake of IPv6

      Given the vitriolic comments I have seen about the design of IPv6 (don't ask me about the validity, I do not know enough about networks), I am not sure if there aren't people who would embrace NDN because it isn't IPv6 and we need something better than IPv4 at some point.

      1. Brewster's Angle Grinder Silver badge

        Re: With incredible uptake of IPv6

        @Michael, I think the OP was being sarcastic.

      2. Christian Berger

        Re: With incredible uptake of IPv6

        Well as usual here, those vitriolic comments come from people who don't understand IP and think NA(P)T is a security feature, just ignore them. Yes IPv6 will have it's initial problems, but it's far from the hell hole IPv4 with carrier grade NAT or NDN will be.

  6. Nigel 11

    Might this be the DAP of networking?

    Before LDAP there was DAP. As far as I know it was never actually fully implemented, let alone ever used on a large scale. It was so encumbered with features and misfeatures, that one might have expected it to sink without trace.

    But someone took a knife to it to create Lightweight DAP and the rest is history. Maybe that history is about to repeat itself. Certainly, the limitations of the internet as it is today are becoming ever more apparent. Something better is needed!

    1. MyffyW Silver badge

      Re: Might this be the DAP of networking?

      @Nigel - maybe. You may recall LDAP itself was considered quite heavy in the mid-nineties, but over time technology (and bandwidth in particular) became available that could handle it.

    2. Suricou Raven

      Re: Might this be the DAP of networking?

      Quite possible. Beneath the monstrosity of NDN lies an older idea, CAN - which is fundamentally brilliant, from a technological point of view. Less so from a business point of view, which is why it hasn't been deployed much - it'd inevitable become a wonderful tool of piracy, and for all the vast expense to ISPs of deploying it they wouldn't be the ones seeing the greatest benefit. Sure, it'd bring the cost of hosting large content down to a fraction of the current price - but it'd also see the ISPs shouldering that fraction, so they obviously have little incentive.

      NDN seems to be an effort to address this by wrapping CAN up in a horrendously over-engineered mess orientated not at actually getting people the data they want, but at rendering the technology more palatable to the business side - giving ISPs a lot more control over the data, so they can do things like ensuring only partnered or reputable businesses can benefit while excluding DodgyFreeMovies.ru and the like.

    3. BitDr

      Re: Might this be the DAP of networking?

      What is needed is better infrastructure, reducing the amount of data flying about it will buy some time but ultimately more capacity is required.

    4. David Roberts

      Re: Might this be the DAP of networking?

      As far as I remember DAP was/is alive and kicking.

      DAP was just too heavyweight (at the time) for client PCs (sound familiar? It is a recurring theme throughout OSI - complex design beyond the capability of budget equipment at the time of design).

      LDAP ran on PCs and connected to an LDAP server which then used DAP as a back end for all the heavy lifting.

  7. David Lawton

    I could be wrong, but if this happens i can see IPv6 never taking off and people will just cling onto IPv4 and jump to this instead when ready.

    1. Yes Me Silver badge

      Yes, you could be wrong...

      > if this happens i can see IPv6 never taking off

      I think you're wrong for 2 reasons:

      1) IPv6 has actually taken off in the last year or so. Just because you don't see it on *your* screen doesn't mean it's not there and growing fast.

      2) NDN is a completely different and *much* more radical change; it won't have any impact on the need for ever more layer 3 addresses, and if it takes off outside the academic world (which is a *big* if) it will co-exist with the non-NDN network for, at a conservative guess, 25 years or more. By which time you will either be using IPv6 or runnning everything via 4 layers of NAT44.

      Don't misunderstand me; I think NDN is a very cool design (and not even remotely comparable to OSI) but it's too radical a change to have an easy start.

      1. Colin Tree

        Re: Yes, you could be wrong...

        Just sounds like object oriented programmers applying themselves to networking. Which means large power hungry routers, switches, NTU's, running buggy, bloated, high level, abstracted code. Maybe they could populate a lot of network caches with a gazillion content addressable memories.

  8. Joe 48

    Am I the only one

    Who spotted the blatant Star Trek quote.

  9. Denarius
    Thumb Down

    X500, X400 and problems

    All I see is another bureaucratic attempt at making something that breaks easily while being impossible to configure. Meanwhile the spooks and Hollywood must be drooling at the thought of something so snooper friendly in concept. UDP 4eva!!

  10. alain williams Silver badge

    Transition hard

    The only way that I can see this coming in is if there is some kind of way of tunneling TCP inside NDN. Yes: this is not what you are supposed to do, but I cannot see people rewriting their applications until a lot of other applications exist -- so a chicken & egg situation.

  11. Anonymous Coward
    Anonymous Coward

    as all future devices will be mobile/wearable

    where is the smart_simple - disruption_tolerant - disruption_expecting - self_mesh_healing comms protocol portion of this particular "Beyond TCP/IP" vision?

    just askin'

  12. Peter Gathercole Silver badge

    I might be being stupid here...

    but... I cannot see anything in the article that suggests the replacement of IP. Indeed, the diagram still has IP listed in layer 2, along with (strangely) UDP. Extrapolating from this, what they may have done is eliminated TCP.

    It looks to me like it is a super-network that sits above the network layer, probably as a way to make it network-independent. It's not in itself going to replace IPv4 or IPv6, which may exist for some time until some other alternative comes along.

    1. Preston Munchensonton

      Re: I might be being stupid here...

      In other words, they've done what everyone else does who needs control and authentication...they wrote an application. All of this bullshit needs to remain safely ensconced in Layers 5-7 where it belongs.

      Personally, I love that TCP/IP doesn't easily provide controls over who can publish content, who can access content, and who can connect to whom. No one needs a system that gives more control to governments and their cronies.

      1. Anonymous Coward
        Anonymous Coward

        Re: I might be being stupid here...

        Eh? You don't see the value of MACsec at layer 2 then, stopping any snooping of your point to point transport links?? Encrypted and authenticated...

        This article is about a protocol that is anti snooping. Anti goverment control. Anti hackable. That's good for those of us who like our data to remain out of Goverments hands. Far less snoopable than tcp/ip and the like. This is a good thing!!!

        1. Charles 9

          Re: I might be being stupid here...

          Perhaps the thought is that a sufficiently-resourced adversary, like a state, can create a Perfect Impersonator and defeat any measure knowable to man.

  13. Anonymous Coward
    Anonymous Coward

    Seems to me that "Security" is an odd thing to have at the 'app' level.

  14. Anonymous Coward
    Anonymous Coward

    <CYNICISM>

    Looks to me like a DRM wet-dream.

    This: “a practically deployable set of protocols replacing TCP/IP that increases network trustworthiness and security, addresses the growing bandwidth requirements of modern content, and simplifies the creation of sophisticated distributed applications.”

    Trustworthiness in this case is not for the end user, it is for the content provider, much as UEFI is not a Trustworthy platform for you and me. Bandwidth is bandwidth, the requirements will grow and the protocol is not going to change that. As for "simplification of distributed applications" again this is not for you and me, its for those selling SAAS to you and me. I don't buy into SAAS, simply because it requires my Internet Access, my Suppliers Internet Access, AND the suppliers servers to be up and running. Remove any one of these and it all goes pear-shaped, as Sales Force's customers are finding out. It also means you MUST have a good high speed connection, and the national infrastructure is simply not there to handle it (last mile problem).

    This: "to preserve “the balance between sharing information and protecting privacy at the same time.”" is simply <COUGH>BULLSH*T</COUGH>. These people are not interested in protecting your privacy, they are interested in getting as much information about you as they can so they can sell it and refine their marketing. Targeting "entities" will help them immensely, especially if you have to register your "entity" (smart TV/Thermostat/Light-bulbs/refrigerator/media-centre/e-book reader/PC/Tablet/Phone etc.).

    When they gush about how the protocol is "trustworthy" they depend on you to think selfishly, i.e. that YOU can trust IT, which our nature predisposes us to do and works to their advantage; sadly it is most likely not the case.

    Then there's this;

    "Crowley goes on to explain that NDN will be able to inform users if data on a bank's site was produced and signed by that bank. IP has no way to perform such a verification. NDN will therefore improve internet security."

    No it does won't improve security, the bank is still ultimately responsible for everything on their site. Connecting to a bank's servers over an encrypted tunnel with a verified secure certificate is ALL that is required. If they screwed up then they owe me. The onus is on the bank.

    Compare this to Chip and PIN technology on debit/credit cards. At no time is the PIN you entered sent across the network, it is all handled internally between the reader and the chip on the card. If that number gets out into the hands of a third party it is either because YOU gave it out or YOU were not diligent in making sure your act of entering the PIN was unobserved. They put the onus on US, the customer, but they sell it as being "more secure"... which it is... from their perspective.

    This: "The consortium says today's internet lacks security because it “ … was designed as a communication network so the only entities that could be named in its packets were communication endpoints.”

    Wow, elegantly simple, a communication network that consists of endpoints and hands off the job of "what to do with the data" to the destination, without a care of what the destination "entity" is.

    Again, security in the context of his statements is from the perspective of those who wish to stream movies sound at your mark-1 eyeballs and ears, sell you books, and so on. It could also be for SAAS or Could Storage, but secure certificates also handle this job rather elegantly. When they describe it as a "more nuanced approach to verifying content's source and connecting entities to content", they're talking about micromanaging who gets what, on what device, when, where and how.

    "The consortium also says its approach won't break the current internet, with the diagram above showing how both offer a “thin waist” allowing communication among many different entities."

    So because the diagram has a thin waist it won't break the Internet? Heck, I can put the diagram in a circle... there... fixed that for ya... using their logic it now breaks the Internet.

    </CYNICISM>

    I feel better.. now... no less worried that the lunatics are in charge of the asylum... but better.

  15. Anonymous Coward
    Anonymous Coward

    It will turn into a kitchen sink effort

    It is inevitable that Big Hollywood and the spooks will come to the table, or send those who get their marching orders from them. They'll get some of what they want, but not all. So they'll hold it back while they work behind the scenes to get the rest of their wishlist.

    Meanwhile, what they did get will cause those who don't want to see DRM or the ability for governments to de-anonymonize dissidents or "suspected terrorists" baked into a new protocol at a low level to rally against this.

    Mark my words, it will never fly.

  16. Anonymous Coward
    Anonymous Coward

    At least they have given it the right name

    In SMTP NDN means a non delivery notification which seems appropriate for this idea which stands no chance of being delivered.

    1. Christian Berger

      Re: At least they have given it the right name

      "idea which stands no chance of being delivered."

      I wouldn't be so positive about that. The advantages for governments and media companies as well as router manufacturers are just to big.

  17. Anonymous Coward
    Facepalm

    The internet lacks security?

    "The consortium says today's internet lacks security because" ..

    No it doesn't, it does exactly what it was designed to do - reliably deliver packets to the endpoints, it's the endpoints that lack security.

    1. Anonymous Coward
      Anonymous Coward

      Re: The internet lacks security?

      You think?

      So tell that to the guy in the cafe getting his http requests redirected in the network/wifi leg...

  18. Christian Berger

    So far, any attempt at such an "intelligent network" have failed

    I mean just look at ATM, X25 and ISDN. All of those networks could in principle do everything we ever wanted, but at a much higher cost. IP took off, because it's so incredibly cheap and easy to implement.

    It also took off since there is no difference between different services. If I wanted to transmit smells, I wouldn't have to go through 10 years of standardisation, I could just transmit the data. And if I do want to have a standard, I just publish my protocols.

    Also there is no concept of "Client" or "Server" in IP. It's in TCP instead. Therefore there is no difference between client and server connections to the Internet, every connection is the same. This is what enabled rapid growth and a vibrant culture.

    An "intelligent network" would turn back the clock to online services like Compuserve or AOL, where you have more or less walled gardens.

    The only problem facing IP(v6) is ISPs which put their customers money into ad campaigns instead of just upgrading their network as they are paid to do. This is why some ISPs have congestion, and this is why we now have widespread outages. The Internet is to important to be left to companies that need to gain the highest profit by law.

  19. smithyBGP

    Lacks commercial understanding of IP Transport Industry

    Universities and research groups can come up with transport and host application session protocols all day long. 99% of these efforts have failed to dent the current usage of BGP and IP on the Internet because the parties concerned do not have a basic understanding of the Commercial implications of operating a network as a business. Your retail, wholesale and international network transport providers, as well as your OTT providers all, rely on the fact that traffic does not always travel on symmetric paths. This happens for a number of reasons and those change depending on the specific provider. In most cases, it's because of congestion, which this protocol conveniently ignores in the routing algorithm.

    " Data packets always take the reverse path of Interests, and, in the absence of packet losses, one Interest packet results in one Data packet on each link, providing flow balance. "

    The simple fact is that no new protocol will take root until it addresses the commercial issues of the infrastructure that it's built on. The replacement cost of that infrastructure is in the Trillions of dollars, and the life cycle to perform that transition would be in the 10's of years. Take IPv6 adoption, that's been going now for ~ 10+ years and still has not made a significant dent in the primary host resolution architecture used today. Commercial workarounds allow businesses to make financial choices that push back on IPv6 as mandatory requirement still to this day.

    As a network infrastructure owner you are not going to write off hundreds of millions of dollars in invested assets, unless there is an evolutionary process that does not result in the write down of the assets and the business model. Network core/distribution/access transport infrastructure does get life cycled on an approximately 3-10 year cycle depending on the role, but that still does not account for the fact that the fiber architectures, DOCSIS infrastructure and other transport mediums, including Wireless are all designed to leverage asymmetric routing to resolve issues related to congestion and cost optimization. Segment Routing, whilst not related to the scope or complexity of Named Data Networking, is one example of an evolution that addresses the evolution of data transport without breaking the fundamental business principals of the network business.

    Named Data Networking as defined, whilst perhaps not targeting the whole of the Layer 1-4 part of the transport problem, does need to address these business issues if it wants to get serious support from the IP networking business community and not just those looking to put their names on research papers and marketing materials required for their next job promotion.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like