Net Neutrality In EU Could Be At Risk With Next Week's Vote

Next week, the European Union Parliament will vote on the net neutrality package as well as other telecom regulations. However, according to activists, the new regulations have some major flaws in them, which could lead to Internet providers abusing their power and ultimately hurting Internet users in the EU.

Latest Videos FromTom's Hardware
Contributor

Lucian Armasu is a Contributing Writer for Tom's Hardware US. He covers software news and the issues surrounding privacy and security.

  • InvalidError
    Network Neutrality sounds nice in theory but it has its downsides as well: some types of services, such as VoIP and competitive gaming, can genuinely benefit from special treatment. Similarly, the IP protocol has provisions for managed QoS and congestion which people and applications could also benefit from if ISPs implemented proper support for them.

    Preventing service providers from experimenting with such new opportunities in the name of network neutrality may ultimately turn into a disservice.

    For example: with real DiffServ QoS, you can floor your connection 24/7 and never worry about running out of bandwidth for VoIP because your VoIP stream has a real-time priority QoS tagging which makes it jump the queue through the network on hops that (try to) honor QoS tags. No need to do ghetto-QoS by reserving a chunk of your typical net bandwidth to make sure you always have at least that much spare bandwidth - at least to the best of your ability from your end of the network.

    Under strict NN rules, such service enhancements which would come a long way towards stabilizing the performance of certain applications during times of heavy network loads become impossible. Seems like it would be a shame to me.

    NN rules may restrict the extent to which "fast lanes", special services, sender-paid/sponsored bandwidth, congestion management, etc. can be used but they should not ban them outright since many of them have very practical legitimate uses.
    Reply
  • skit75
    ....Just like toll lanes improve automobile traffic?

    Oversimplifying, you bet. The toll lanes sounded good in theory and study after study says they have only made traffic worse, evidenced by 6 lanes of bumper to bumper traffic on the 405 or 110 while two full lanes go nearly unused.

    Let the fast lane proponents fund and build their own lanes if that is what they want. Keep the ISP out of it. We need all of our lanes even if we don't use all of them, all of the time. When we need them, we REALLY need them.
    Reply
  • InvalidError
    16830913 said:
    Let the fast lane proponents fund and build their own lanes if that is what they want. Keep the ISP out of it. We need all of our lanes even if we don't use all of them, all of the time. When we need them, we REALLY need them.
    The proper way to go about it would be to have companies, ISPs, service providers and other network operators who want fast lanes to build them but then require them to make whatever under-used capacity there is on it available for all other traffic. This way, the prioritized/fast-lane traffic is always fast and other lanes get some relief when the fast lanes are not full. This way, even the "toll lanes" get packed during rush hours and people who paid for the toll lanes still get to spare their brakes.
    Reply
  • 3dhoc
    The "fast lanes" sound nice in theory, but are they really needed? I've got just a DSL connection and stuff works JUST FINE. Plus, letting the government decide anything about our information during such political crisis is a bad idea. In the end, the ISPs probably just figured that they are too lazy to actually try offer such a service and went directly for the parliament to FORCE us in such a service. If I want fast lanes, I want it to be in my contract. I paid for a plain connection that should allow all services to work. Instead of wasting our time on QoS we should be researching and installing new lanes... QoS was created during the times of modems and barely makes sense now, except for specialized scenarios. At my home I do not want it, do not need it.
    Reply
  • InvalidError
    16832229 said:
    Instead of wasting our time on QoS we should be researching and installing new lanes... QoS was created during the times of modems and barely makes sense now, except for specialized scenarios.
    QoS capabilities are already built into all network equipment more advanced than SOHO router and switch. It has been used by networks and service providers for to guarantee the performance of certain traffic, such as phone calls, over mixed networks for decades in one form or another.

    Adding capacity is not enough to guarantee the performance of a timing-critical stream over the internet: internet traffic comes and goes in bursts. It is impossible to guarantee that no buffers anywhere between A and B won't get backed up by a burst even if you have 10X more capacity than normally needed. It only takes a few milliseconds of jitter to severely degrade the quality of a VoIP call.

    Brute force can only get you so far and gets real expensive real fast. With QoS, networks could work smarter instead of harder.

    QoS is not really that specialized: if you have FTTH, the service provider relies on QoS to make IPTV, voice and data play nice on the shared fiber. Without QoS on the fiber, they would be unable to guarantee POTS-equivalent call quality.
    Reply
  • 3dhoc
    NOTE: Tom's, the commenting usability is abysmal. I just double-posted while being totally blinded by the wind from scrolling around while looking for buttons to do simple things.

    Now, I know the technicalities. Still, thanks for the Informatics lesson. IPTV you say? Are you serious? I want all my streams encrypted. QoS only works if packets can be analyzed. This legislation is scary, because it would allow my privacy to be de-prioritized amongst various other dumb services. The parliament is wasting our time and money.
    Reply
  • InvalidError
    16833634 said:
    I want all my streams encrypted. QoS only works if packets can be analyzed. This legislation is scary, because it would allow my privacy to be de-prioritized amongst various other dumb services.
    DiffServ QoS does not need to know anything about your packets other than the QoS field in the IP headers and (optionally) source/destination IPs if the network needs to enforce use of a specific QoS tag for a specific service or source/destination. Everything QoS-based forwarding absolutely needs is part of the basic IP headers. In older versions of the IPv4 header definition, it used to be called Differentiated Service Code Point or DSCP.

    How do you do blind QoS? You assign bit buckets, peak and sustained data rates and a priority level to each supported QoS tag (managed switches, enterprise/carrier-grade routers and converged network modems/ONTs already have all of this built-in) and then all you need to do is tell your application to set the DiffServ field to the appropriate value. For networks that follow the recommended DiffServ definitions, VoIP traffic for example would typically be tagged with code 46 (x2E) which should correspond to Expedited Forwarding. If you ever configured a VoIP device, you may have noticed an option to set the DiffServ field that defaults to 46, that's what it is for.

    If you try pushing more bits through a given QoS channel than what you or your upstream provider(s) paid your ISP for and your ISP set your limits at, your QoS bit bucket for that QoS tag dries up and congests, preventing you from using the QoS channel for anything that uses more bandwidth than what it has been provided and intended for. You can VPN through your 200kbps VoIP QoS channel all day if you want.
    Reply
  • Sakkura
    The author doesn't seem to be aware that roaming applies to data as well as calls. Data roaming will remain important, in fact it will become a bigger issue over time.
    Reply
  • QSV
    Network Neutrality sounds nice in theory but it has its downsides as well: some types of services, such as VoIP and competitive gaming, can genuinely benefit from special treatment. Similarly, the IP protocol has provisions for managed QoS and congestion which people and applications could also benefit from if ISPs implemented proper support for them.

    Preventing service providers from experimenting with such new opportunities in the name of network neutrality may ultimately turn into a disservice.

    For example: with real DiffServ QoS, you can floor your connection 24/7 and never worry about running out of bandwidth for VoIP because your VoIP stream has a real-time priority QoS tagging which makes it jump the queue through the network on hops that (try to) honor QoS tags. No need to do ghetto-QoS by reserving a chunk of your typical net bandwidth to make sure you always have at least that much spare bandwidth - at least to the best of your ability from your end of the network.

    Under strict NN rules, such service enhancements which would come a long way towards stabilizing the performance of certain applications during times of heavy network loads become impossible. Seems like it would be a shame to me.

    NN rules may restrict the extent to which "fast lanes", special services, sender-paid/sponsored bandwidth, congestion management, etc. can be used but they should not ban them outright since many of them have very practical legitimate uses.

    Its funny how few people really understand net neutrality. Even here. Its really simple: It means that it is as fast as possible in any case, no matter what you do with your connection. Youre starting at a base that has no net neutrality!
    Reply
  • InvalidError
    16837500 said:
    Its really simple: It means that it is as fast as possible in any case, no matter what you do with your connection.
    Being fast is one thing, being predictably and reliably fast are another.

    Even with a proper DiffServ PHB setup, you still need 3-5X the QoS'd bandwidth to guarantee with reasonable certainty that there won't be much priority traffic backlogs (buffering) when packets from multiple QoS'd streams try to cross the same links between routers at the same time.

    Building the whole internet in such a way that all traffic can be treated as priority would be prohibitively expensive.
    Reply