And then, for the nth time, RDP, gaming, security cameras, and lots of other services won't work correctly either, because they require more bandwidth and latency that is at least as low as VoIP.
And for the nth time, you haven't addressed the issue - residential bandwidth is not committed. If its not committed, how do you guarantee the priority traffic over non-priority traffic when the available traffic fluctuates?
RDP, gaming and security cameras may or may not have higher real-time requirements, but I would argue that the user of these applications is more willing to accept issues caused by network fluctuations. RDP and security cameras especially. see my earlier arguments why gaming and voip aren't comparable from a usage perspective.
With a decent codec, VoIP requires less than 8kbps, and it pretty much doesn't matter whether you use TCP, UDP, or some other network protocol. Using G.711 for VoIP is just plain stupid
G.729 @ 20ms frames is 24kbps with UDP/IP overhead. G.711 runs a majority of all Voip today. I agree that other codecs will take over, however, I think they'll be closer to 40kbps with overhead to deliver MOS scores closer to 4.0. In the end, it doesn't matter. Its not an issue of bandwidth or latency, but an issue of guaranteed delivery.
I pointed out the possibility that real-time traffic flows through a VPN, you point out the possibility that a VPN might not carry real-time traffic. My point is relevant to the discussion, yours is not: if ISPs support VPNs (and they have to), they have to support them well enough so that the real-time traffic that may flow through those VPNs arrives in time at the other end.
In fact, any use of VPN is
No, you didn't point that out. You made the claim that any use of VPN is.
Although I've enjoyed this conversation, let's call it quits. You haven't addressed my argument other than to call it far-fetched with no argument as to why it is. Your argument centers around claiming that the requirements of other traffic flows are higher than voip, which I don't dispute. I've spent several years deploying voip on networks, from the largest carriers to a residential ISP. I'm basing my argument on true network evidence and experience. VOIP works and is definately going to change the market. But for residential usage, the nature of an oversubscribed network gives the advantage to the entity that owns the oversubscribed network.
Let's call it quits and let the market prove out who is correct. If your in the denver area, we could wager a beer on how it shakes out if you like. I'm always interested in meeting fellow passionate tech junkies.
VPNs have the real-time requirements of the traffic that runs through them.
Agreed. At the base, maintaining a VPN connection does not require anywhere near the real-time capabilities of voice. We are both guilty of making blanket statements about VPN traffic patterns.
You sound like a nerd living at home by himself. In fact, ISPs better offer reasonably good simultaneous connections, because otherwise, in a normal family, junior (playing games), dad (on RDP to the office), mom (listening to MP3s), and sister (downloading an ISO) are going to be pissed. Or, for that matter, are multiple small office users.
I tried to make my example simpistic so you would understand. Adding all those simultaneous uses increases the likely hood of it occuring. I would probably keep your personal attacks to yourself, especially if you haven't thought through your argument.
Sorry, even your far fetched scenario isn't realistic. Besides, you are an order of magnitude off on the requirements of voice.
Far fetched. Let me see. Simultaneous use of internet, plus real-time voice traffic, plus network traffic node usage fluctuations. Sounds like the last mile network of most major residential ISP. Order of magnitude off - go back and read up on voip. G.711 - 80kbps for 20ms frames when you add UDP/IP packet overhead. So 80*20 = 1600kps. Oh, I'm sorry, the ISP is delivering 30X the requirements for a single call and still has possibility that the call quality could be affected by node traffic fluctuations.
When you've deployed voip on an ISPs network, managed the traffic utilization at aggregation points, and witnessed what happens when end-user traffic rapidly exceeds the node capacity you'll realize that the ISP is not what you imagine it to be - massive amounts of bandwidth dedicated to each end user. Again, the game is oversubscription, and in that game, the last mile provider has the extra knowledge the other providers don't have. I'm not saying broadband agnostic voice won't work, I'm saying that the last mile providers will provide a solution that exceeds PSTN at some point and the other providers will struggle to match it.
than VoIP itself. Gaming is a major application (in fact, voice in gaming is a major application) that requires stricter real-time guarantees than VoIP and more bandwidth. In fact, any use of VPN is
What it is in real life is something whose average performance exceeds the needs of VoIP manyfold and across which more and more content runs encrypted and opaque.
And what I am saying is that any broadband ISP that doesn't achieve >80kbps and
Surely your kidding - VPNs don't have more strict real-time requirements than voice. It doesn't matter. The issue of degredation is two fold. Premise bandwidth delivered and backbone bandwidth available. You are confusing the ability of an ISP to reliably deliver 80kbps to the premise and the ability of the ISP to deliver 80kbps of real-time voice traffic when the user is performing other tasks. If I'm a gamer, I'm not pulling an ISO at the same time that I'm playing - why? - because it affects my latency/lag. If I'm a voice user expecting to make a call, I'm not worried about my ISP download - why? - because "I can be on the Internet and the phone at the same time".
Lets say the user is downloading at 3mbps (his promised speed). He starts a voip call. His ATA/Router (w/ traffic shaper set to a download of 2800kbps to protect his call) properly protects his download from causing voice issues. But, the node he sits on gets slammed with traffic usage. Suddenly, the ISP is delivering 2500mbps to the user. Now his download can affect his voice call. So the ISP is delivering 20X the requirements for a basic call and yet the end user can experience a call degredation.
Yes, your ATA/Router could begin to detect the jitter/latency changes in real-time and try to compensate. However this will be after the initial issue occurs and the complexity will raise the price of the CPE.
You have the same blinders on as telcos. You think of the Internet as a hugely unreliable, slow network running a bunch of transparent protocols. What it is in real life is something whose average performance exceeds the needs of VoIP manyfold and across which more and more content runs encrypted and opaque
Nope, I see the Internet as an interconnected group of reliable, flexible networks. But the last mile provider has an advantage as shown above. It doesn't matter how opaque the traffic is if the network is oversubscribed - the last mile provider can compensate for the oversubscription while the other providers can't.
How does this get modded insightful? Sure, on a network without oversubscription, you don't need to prioritize traffic. However, the entire premise of ISPs is oversubscription, especially for residential service. Sure ISPs like Speakeasy promise speed guarantees - that's easy, they just deliver full speed a majority of the time. So, if some of the data gets delayed during a small window, who's to know better? With voice that degradation causes packet drops as the jitter buffers are exceeded.
What TFA is saying is that the network operators will prioritize their own voice so that it always gets delivered. Each line requires only 80kbps for G.711 (I know, G.729 uses less bandwidth, still the same issue holds). Guaranteeing 80kbps to every household is doable as opposed to 4mbps ( if you think the cable company is committed to delivering you 4mpbs, you are smoking crack). So their voice traffic gets a guarantee and the alternative products are left susceptible to node/dslam usage fluctuations.
As the market continues to move from early adopters to early majority, the quality expectation is going to continue to rise (no more outages and toll-quality voice). Will an end user pay $5-10 more per month for their ISPs own service if the quality is better?
What do the Vonages of the world do? Ask for the FCC to force QOS regulation? Please, they asked for an unregulated industry and even if they were able to get some sort of regulation, the cable/ILEC guys will butcher them with regulation gamesmanship (just ask the CLECs today). Nope, I would guess that they will seek to work with the biggest providers to pay for QOS guarantees or try to figure out their exit plan.
And to the point that service/application based filtering is techinically hard - nope. There are plenty of boxes that do this. I've used them to prioritize voice (tracks the sip signalling and matches up the RTP associated with the call) . Application based filtering is a given. Check out netintact.com for of just one solution already deployed.
The ISPs don't need to port block. They can block at the application layer. Numerous solutions exist out there to shape traffic at the application layer, including recognizing the RTP traffic associated with a SIP call.
I don't think the big guys will block VOIP. They don't need to. If they prioritize their in-house VOIP traffic and then just treat the BYOV (Bring Your Own Voip) traffic as normal web - or perhaps even a lower level, the call experience will not be as good on the BYOV as the in-house. Bundle this with their marketing power, triple play capabilities, and power backup (the cable companies are already building out power backup), and they won't need to block it - the "market" will work it out for them.
How is this a negative? HP is stepping forward and resolving any customer concerns about running linux on HP systems. The more companies that follow suit, the less revenue SCO will be able to generate from linux licensing fees. And publicly it makes it clear that HP does not think SCO has a case.
Comparing HP to SCO is ridiculous. Would you have them assume responsibility for every linux user? No CFO in their right mind would assume that risk for no possible return. Put aside your "all corporations are evil" fanaticism and see this for what it is - a positive move by a large equipment vendor that will help dismiss the scare tactics taken by SCO.
Slashdot Mirror
And then, for the nth time, RDP, gaming, security cameras, and lots of other services won't work correctly either, because they require more bandwidth and latency that is at least as low as VoIP.
And for the nth time, you haven't addressed the issue - residential bandwidth is not committed. If its not committed, how do you guarantee the priority traffic over non-priority traffic when the available traffic fluctuates?
RDP, gaming and security cameras may or may not have higher real-time requirements, but I would argue that the user of these applications is more willing to accept issues caused by network fluctuations. RDP and security cameras especially. see my earlier arguments why gaming and voip aren't comparable from a usage perspective.
With a decent codec, VoIP requires less than 8kbps, and it pretty much doesn't matter whether you use TCP, UDP, or some other network protocol. Using G.711 for VoIP is just plain stupid
G.729 @ 20ms frames is 24kbps with UDP/IP overhead. G.711 runs a majority of all Voip today. I agree that other codecs will take over, however, I think they'll be closer to 40kbps with overhead to deliver MOS scores closer to 4.0. In the end, it doesn't matter. Its not an issue of bandwidth or latency, but an issue of guaranteed delivery.
I pointed out the possibility that real-time traffic flows through a VPN, you point out the possibility that a VPN might not carry real-time traffic. My point is relevant to the discussion, yours is not: if ISPs support VPNs (and they have to), they have to support them well enough so that the real-time traffic that may flow through those VPNs arrives in time at the other end.
In fact, any use of VPN is
No, you didn't point that out. You made the claim that any use of VPN is.
Although I've enjoyed this conversation, let's call it quits. You haven't addressed my argument other than to call it far-fetched with no argument as to why it is. Your argument centers around claiming that the requirements of other traffic flows are higher than voip, which I don't dispute. I've spent several years deploying voip on networks, from the largest carriers to a residential ISP. I'm basing my argument on true network evidence and experience. VOIP works and is definately going to change the market. But for residential usage, the nature of an oversubscribed network gives the advantage to the entity that owns the oversubscribed network.
Let's call it quits and let the market prove out who is correct. If your in the denver area, we could wager a beer on how it shakes out if you like. I'm always interested in meeting fellow passionate tech junkies.
VPNs have the real-time requirements of the traffic that runs through them.
Agreed. At the base, maintaining a VPN connection does not require anywhere near the real-time capabilities of voice. We are both guilty of making blanket statements about VPN traffic patterns.
You sound like a nerd living at home by himself. In fact, ISPs better offer reasonably good simultaneous connections, because otherwise, in a normal family, junior (playing games), dad (on RDP to the office), mom (listening to MP3s), and sister (downloading an ISO) are going to be pissed. Or, for that matter, are multiple small office users.
I tried to make my example simpistic so you would understand. Adding all those simultaneous uses increases the likely hood of it occuring. I would probably keep your personal attacks to yourself, especially if you haven't thought through your argument.
Sorry, even your far fetched scenario isn't realistic. Besides, you are an order of magnitude off on the requirements of voice.
Far fetched. Let me see. Simultaneous use of internet, plus real-time voice traffic, plus network traffic node usage fluctuations. Sounds like the last mile network of most major residential ISP. Order of magnitude off - go back and read up on voip. G.711 - 80kbps for 20ms frames when you add UDP/IP packet overhead. So 80*20 = 1600kps. Oh, I'm sorry, the ISP is delivering 30X the requirements for a single call and still has possibility that the call quality could be affected by node traffic fluctuations.
When you've deployed voip on an ISPs network, managed the traffic utilization at aggregation points, and witnessed what happens when end-user traffic rapidly exceeds the node capacity you'll realize that the ISP is not what you imagine it to be - massive amounts of bandwidth dedicated to each end user. Again, the game is oversubscription, and in that game, the last mile provider has the extra knowledge the other providers don't have. I'm not saying broadband agnostic voice won't work, I'm saying that the last mile providers will provide a solution that exceeds PSTN at some point and the other providers will struggle to match it.
than VoIP itself. Gaming is a major application (in fact, voice in gaming is a major application) that requires stricter real-time guarantees than VoIP and more bandwidth. In fact, any use of VPN is
What it is in real life is something whose average performance exceeds the needs of VoIP manyfold and across which more and more content runs encrypted and opaque.
And what I am saying is that any broadband ISP that doesn't achieve >80kbps and
Surely your kidding - VPNs don't have more strict real-time requirements than voice. It doesn't matter. The issue of degredation is two fold. Premise bandwidth delivered and backbone bandwidth available. You are confusing the ability of an ISP to reliably deliver 80kbps to the premise and the ability of the ISP to deliver 80kbps of real-time voice traffic when the user is performing other tasks. If I'm a gamer, I'm not pulling an ISO at the same time that I'm playing - why? - because it affects my latency/lag. If I'm a voice user expecting to make a call, I'm not worried about my ISP download - why? - because "I can be on the Internet and the phone at the same time".
Lets say the user is downloading at 3mbps (his promised speed). He starts a voip call. His ATA/Router (w/ traffic shaper set to a download of 2800kbps to protect his call) properly protects his download from causing voice issues. But, the node he sits on gets slammed with traffic usage. Suddenly, the ISP is delivering 2500mbps to the user. Now his download can affect his voice call. So the ISP is delivering 20X the requirements for a basic call and yet the end user can experience a call degredation.
Yes, your ATA/Router could begin to detect the jitter/latency changes in real-time and try to compensate. However this will be after the initial issue occurs and the complexity will raise the price of the CPE.
You have the same blinders on as telcos. You think of the Internet as a hugely unreliable, slow network running a bunch of transparent protocols. What it is in real life is something whose average performance exceeds the needs of VoIP manyfold and across which more and more content runs encrypted and opaque
Nope, I see the Internet as an interconnected group of reliable, flexible networks. But the last mile provider has an advantage as shown above. It doesn't matter how opaque the traffic is if the network is oversubscribed - the last mile provider can compensate for the oversubscription while the other providers can't.
How does this get modded insightful? Sure, on a network without oversubscription, you don't need to prioritize traffic. However, the entire premise of ISPs is oversubscription, especially for residential service. Sure ISPs like Speakeasy promise speed guarantees - that's easy, they just deliver full speed a majority of the time. So, if some of the data gets delayed during a small window, who's to know better? With voice that degradation causes packet drops as the jitter buffers are exceeded.
What TFA is saying is that the network operators will prioritize their own voice so that it always gets delivered. Each line requires only 80kbps for G.711 (I know, G.729 uses less bandwidth, still the same issue holds). Guaranteeing 80kbps to every household is doable as opposed to 4mbps ( if you think the cable company is committed to delivering you 4mpbs, you are smoking crack). So their voice traffic gets a guarantee and the alternative products are left susceptible to node/dslam usage fluctuations.
As the market continues to move from early adopters to early majority, the quality expectation is going to continue to rise (no more outages and toll-quality voice). Will an end user pay $5-10 more per month for their ISPs own service if the quality is better?
What do the Vonages of the world do? Ask for the FCC to force QOS regulation? Please, they asked for an unregulated industry and even if they were able to get some sort of regulation, the cable/ILEC guys will butcher them with regulation gamesmanship (just ask the CLECs today). Nope, I would guess that they will seek to work with the biggest providers to pay for QOS guarantees or try to figure out their exit plan.
And to the point that service/application based filtering is techinically hard - nope. There are plenty of boxes that do this. I've used them to prioritize voice (tracks the sip signalling and matches up the RTP associated with the call) . Application based filtering is a given. Check out netintact.com for of just one solution already deployed.
The ISPs don't need to port block. They can block at the application layer. Numerous solutions exist out there to shape traffic at the application layer, including recognizing the RTP traffic associated with a SIP call.
I don't think the big guys will block VOIP. They don't need to. If they prioritize their in-house VOIP traffic and then just treat the BYOV (Bring Your Own Voip) traffic as normal web - or perhaps even a lower level, the call experience will not be as good on the BYOV as the in-house. Bundle this with their marketing power, triple play capabilities, and power backup (the cable companies are already building out power backup), and they won't need to block it - the "market" will work it out for them.
How is this a negative? HP is stepping forward and resolving any customer concerns about running linux on HP systems. The more companies that follow suit, the less revenue SCO will be able to generate from linux licensing fees. And publicly it makes it clear that HP does not think SCO has a case.
Comparing HP to SCO is ridiculous. Would you have them assume responsibility for every linux user? No CFO in their right mind would assume that risk for no possible return. Put aside your "all corporations are evil" fanaticism and see this for what it is - a positive move by a large equipment vendor that will help dismiss the scare tactics taken by SCO.