Net Neutrality

[Palerider]

Mark Cuban ...owner of the Dallas Mavericks is very strong opinion on the matter....I would share the link but my pad won't allow me too. I guess it doesn't like the link. Lol

[Guest]

You think so?

Out of curiosity, how do you define network neutrality? What would you like to see ISPs doing with the traffic going through their networks?

 

Yes.  Very much so.  As a principle of ideal conditions.

 

Are you serious about me defining net neutrality?  Or was that rhetorical?

 

ISP's doing with the traffic going through their networks?  Simple - send it from point A to point B and let the end-points make the decisions.

[Mahone]

Yes. Very much so. As a principle of ideal conditions.

Are you serious about me defining net neutrality? Or was that rhetorical?

ISP's doing with the traffic going through their networks? Simple - send it from point A to point B and let the end-points make the decisions.

Yes, I was serious. Mainly because everyone is talking about net neutrality, but everyone seems to have a slightly different idea of what it actually entails and how it will be done.

Personally I like the idea of all traffic treated equally as a basic principle, but it leaves a few questions that need answering. For example, what about quality of service rules for traffic such as VoIP? Controlling that from the end point doesn't help much. The point of QoS is that higher priority QoS is given a higher priority throughout the life of the packet, not just at the end points.

VoIP traffic needs low latency to work. For instance, if VoIP traffic gets stuck behind facebook HTTP traffic, the VoIP users are going to notice very quickly when the conversation starts breaking up.

On the flip side, the Facebook user is unlikely to notice any difference if they have to wait a further split second for their Facebook chat message to send because the VoIP traffic was prioritized.

The basic principle as you gave it will allow this problem to continue, and no-one will be allowed to try and combat it.

Edited February 27, 2015 by Mahone

[Guest]

Yes, I was serious. Mainly because everyone is talking about net neutrality, but everyone seems to have a slightly different idea of what it actually entails and how it will be done.

Personally I like the idea of all traffic treated equally as a basic principle, but it leaves a few questions that need answering. For example, what about quality of service rules for traffic such as VoIP? Controlling that from the end point doesn't help much. The point of QoS is that higher priority QoS is given a higher priority throughout the life of the packet, not just at the end points.

VoIP traffic needs low latency to work. For instance, if VoIP traffic gets stuck behind facebook HTTP traffic, the VoIP users are going to notice very quickly when the conversation starts breaking up.

On the flip side, the Facebook user is unlikely to notice any difference if they have to wait a further split second for their Facebook chat message to send because the VoIP traffic was prioritized.

The basic principle as you gave it will allow this problem to continue, and no-one will be allowed to try and combat it.

 

The VoIP versus FB... or more likely small data two-way type (telephony) versus big data one-way type (streaming audio/video) and the regular data packets of web traffic doesn't need to be solved by isp discrimination... I see it like how the Freeway doesn't solve the problem of vehicular traffic when one has to be at work at 8AM versus the next car over who is enjoying a vacation.  But... the Freeway may have specific lanes assigned to high-occupancy vehicles - and all hovs and only hovs use this lane.  On the internet - QoS can be addressed through protocol and data layers. The ISPs still send packets indiscriminately from point A to point B... the end systems determine via protocols when the data packet enters the lane and which lane it uses.  ISPs may have fast lanes and slow lanes... but anything - regardless of content - on the fast lane gets multiplexed faster than the slow lane.  So that ISPs still maintain net neutrality while end-systems (or as close to the end as is technically feasible) control how the packets line up.

 

Make sense?

Edited February 27, 2015 by anatess

[Guest]

In any case... net neutrality should remain the responsibility of CIS systems and not the FCC... with solutions brought through innovation and capitalist principles rather than government regulation.

[Mahone]

The VoIP versus FB... or more likely small data two-way type (telephony) versus big data one-way type (streaming audio/video) and the regular data packets of web traffic doesn't need to be solved by isp discrimination... I see it like how the Freeway doesn't solve the problem of vehicular traffic when one has to be at work at 8AM versus the next car over who is enjoying a vacation. But... the Freeway may have specific lanes assigned to high-occupancy vehicles - and all hovs and only hovs use this lane. On the internet - QoS can be addressed through protocol and data layers. The ISPs still send packets indiscriminately from point A to point B... the end systems determine via protocols when the data packet enters the lane and which lane it uses. ISPs may have fast lanes and slow lanes... but anything - regardless of content - on the fast lane gets multiplexed faster than the slow lane. So that ISPs still maintain net neutrality while end-systems (or as close to the end as is technically feasible) control how the packets line up.

Make sense?

As far as I can tell, your solution is at the physical layer. Are you proposing that ISPs have two separate physical infrastructures, one which is the "fast lane" and offers high speed low latency service for all traffic send down it, and another which offers a lower quality connection?

Just trying to clarify.

Edited February 28, 2015 by Mahone

[jerome1232]

I can't remember the details exactly but they do offer a service that bypasses the internet backbone and ties directly into the isp's network. At least it is something to that effect.

[Mahone]

I can't remember the details exactly but they do offer a service that bypasses the internet backbone and ties directly into the isp's network. At least it is something to that effect.

I assume you're referring to very large service providers such as netflix?

[jerome1232]

Again, as I understand it, to anyone willing to pay for it. As I understand it services like Netflix which deliver large amounts of content which requires good quality of service depend on these arrangements to deliver.

[Mahone]

Again, as I understand it, to anyone willing to pay for it. As I understand it services like Netflix which deliver large amounts of content which requires good quality of service depend on these arrangements to deliver.

Yes, my understanding is that the Internet Service Providers are expected by netflix to pay for the additional infrastructure to connect the two networks directly, as opposed to routing through a higher tier network.

If an ISP refuses to cough up, netflix traffic to their customers will likely suffer from additional delays and slow speeds.

[jerome1232]

I wasn't aware of that angle. Is that the pro-"net neutrality" position, that an ISP must take the hit?

[Palerider]

http://www.newsmax.com/TheWire/seven-reasons-net-neutrality-threat/2015/02/27/id/627247/

good read

[bytebear]

Hey, my grocery store may stop providing bananas, therefore, we must regulate grocery stores.   If anyone thinks this is about protecting consumers, and not a power grab, is a fool.

[Guest]

As far as I can tell, your solution is at the physical layer. Are you proposing that ISPs have two separate physical infrastructures, one which is the "fast lane" and offers high speed low latency service for all traffic send down it, and another which offers a lower quality connection?

Just trying to clarify.

 

Not necessarily although it is, of course, a part of the solution.  Besides the physical layer, this can also be done through the data layer and establishment of certain protocols that affect sending multiplexers but removing that process out of the ISP's systems.  Basically... in net neutrality - data is data (faceless) once it hits the ISP line... but end-systems can be designed to carry the burden of prioritization rather than the ISPs.

 

And this is why I find it very important to keep the government out of net neutrality... because... we are now opening the door for the government to insert themselves between the end systems and the ISP... with the authority to dictate how the data is going to be choked.

Edited March 2, 2015 by anatess

[Mahone]

Not necessarily although it is, of course, a part of the solution. Besides the physical layer, this can also be done through the data layer and establishment of certain protocols that affect sending multiplexers but removing that process out of the ISP's systems. Basically... in net neutrality - data is data (faceless) once it hits the ISP line... but end-systems can be designed to carry the burden of prioritization rather than the ISPs

I believe this proposal will end up costing end users considerably more.

So let's assume the statistical multiplexing that the ISP performs on their physical links is modified to prioritize all and any traffic that originated from or is going to the "premium" end user connections. This also includes traffic that is really unnecessary to prioritize, such as torrent downloading, youtube videos and any real time one way streaming that can buffer, just because the customer is saying it should be, by pushing it through that premium link.

At some point, during peak times, that premium traffic is going to exceed the throughput that the ISPs equipment can handle and things start slowing down. Users with services like VoIP will immediately begin to notice, as this service requires low latency.

Currently ISPs can perform deep packet inspection and have the option of sticking a QoS tag onto the data protocol unit to prioritize the traffic for services that will really notice the delay, such as VoIP. Anything not prioritized is unlikely to notice the delay of only a few milliseconds, such as standard HTTP traffic. This method can avoid having to upgrade expensive equipment and install additional physical links between PoPs, the cost of which would ultimately get passed onto the consumers.

Edited March 3, 2015 by Mahone

[Mahone]

I wasn't aware of that angle. Is that the pro-"net neutrality" position, that an ISP must take the hit?

I'm not sure it has a position with net neutrality, as technically the traffic is treated equally still, just more efficient routing. Looking at the news articles since I first read about it, I think despite netflix claims that the ISPs should pay, netflix ending up paying for direct physical links to the ISPs networks in question.

Edited March 3, 2015 by Mahone

[Guest]

I believe this proposal will end up costing end users considerably more.

So let's assume the statistical multiplexing that the ISP performs on their physical links is modified to prioritize all and any traffic that originated from or is going to the "premium" end user connections. This also includes traffic that is really unnecessary to prioritize, such as torrent downloading, youtube videos and any real time one way streaming that can buffer, just because the customer is saying it should be, by pushing it through that premium link.

At some point, during peak times, that premium traffic is going to exceed the throughput that the ISPs equipment can handle and things start slowing down. Users with services like VoIP will immediately begin to notice, as this service requires low latency.

Currently ISPs can perform deep packet inspection and have the option of sticking a QoS tag onto the data protocol unit to prioritize the traffic for services that will really notice the delay, such as VoIP. Anything not prioritized is unlikely to notice the delay of only a few milliseconds, such as standard HTTP traffic. This method can avoid having to upgrade expensive equipment and install additional physical links between PoPs, the cost of which would ultimately get passed onto the consumers.

 

We're basically saying the same thing except... Instead of the ISPs doing it, end systems are doing it (or as close to the end as possible) where its only job is to queue data.

 

Youtube/Hulu/AppleTV/movie theaters (one way big data streaming) and VoIP/gaming (two way relatively smaller low-latency data punches) are two different things requiring two different protocols.  Facetime-like communication is also another different thing (two way), relatively bigger data streaming).

 

So this prioritization is done outside of the ISP.  My suggestion is to have things like the IEEE or some commercial entity come up with required QoS protocols for the queue to cover that "last mile" issue and be somewhat of a Gatekeeper even incorporating tiered cost structures for it.  Any new innovations on cabling/wireless comms that provide fastest data transfers can be incorporated into Gatekeeping but it is not under the ISP's umbrella... so that the Gatekeeping process can be made more competitive than the existing ISP-wars.

 

Yes, this could potentially remove the incentive for ISPs to innovate, but I don't see this as a big issue because as it is right now we're still in the "wonder" age where we are still pushing to be Star Trekky for the sole purpose of seeing if we can.

[Vort]

Yes, this could potentially remove the incentive for ISPs to innovate, but I don't see this as a big issue because as it is right now we're still in the "wonder" age where we are still pushing to be Star Trekky for the sole purpose of seeing if we can.

Funny. I have suspected this for years, but I've never heard anyone actually give voice to the idea.

[Mahone]

We're basically saying the same thing except... Instead of the ISPs doing it, end systems are doing it (or as close to the end as possible) where its only job is to queue data.

 

Youtube/Hulu/AppleTV/movie theaters (one way big data streaming) and VoIP/gaming (two way relatively smaller low-latency data punches) are two different things requiring two different protocols.  Facetime-like communication is also another different thing (two way), relatively bigger data streaming).

 

So this prioritization is done outside of the ISP.  My suggestion is to have things like the IEEE or some commercial entity come up with required QoS protocols for the queue to cover that "last mile" issue and be somewhat of a Gatekeeper even incorporating tiered cost structures for it.  Any new innovations on cabling/wireless comms that provide fastest data transfers can be incorporated into Gatekeeping but it is not under the ISP's umbrella... so that the Gatekeeping process can be made more competitive than the existing ISP-wars.

 

Yes, this could potentially remove the incentive for ISPs to innovate, but I don't see this as a big issue because as it is right now we're still in the "wonder" age where we are still pushing to be Star Trekky for the sole purpose of seeing if we can.

 

We are partly saying the same thing, but it isn't just a "last mile" issue (unless by last mile you're referring to the whole of the ISPs infrastructure - such terminology has multiple meanings in the networking world). It's across the entire ISPs infrastructure, and beyond that to the higher tier networks and internet backbones - but how it works once the traffic leaves the ISPs network is a different discussion. I'd agree that a third party would be a good compromise on deciding which data gets which QoS rules, but those rules need to apply right the way through the ISPs infrastructure. Essentially, I don't believe all traffic should be treated equally by the ISP, I think QoS should be implemented if required. But I'd be happy for a third party to decide how those QoS rules are applied.

 

Bear in mind that I'm not saying this technology requires implementing widely at present on ISPs networks, but I do think the ISPs should be able to resort to this if necessary.

Edited March 7, 2015 by Mahone