On Jun 6, 2014, at 7:30 PM, "Manfredi, Albert E" <albert.e.manfredi@xxxxxxxxxx> wrote: > > I think you need to look up how these ISP networks are architected, Craig. > The ADSL link from CO to homes is just the very last leg. The core ISP > network is the main event, and it's DEFINITELY shared for all manner of > different VPNs. There are very many techniques available to achieve this. You > can bet that Verizon does not deploy individual physical infrastructures for > every client intranet or domestic broadband access it needs to provide. > > Read this. And note that MPLS (multiprotocol label switching) is merely one > of many techniques used for deploying VPNs over a common infrastructure. > > http://www.cisco.com/c/en/us/solutions/collateral/enterprise/managed-services-enterprise/net_customer_profile0900aecd80234ba3.html Thanks for the reference. Clearly there is a lot going on in the CO and the WAN that can impact QOS, not just the last mile. The article you linked to identified the challenges, and noted that the primary way they were overcome was: "A QoS-enabled, high-bandwidth network met this challenge" Interesting words, but what do they mean by QOS enabled? > The main point in this is that "fast lanes," or "priority lanes," or whatever > you want to call them, are a perfectly legitimate way of deploying networks. > The FCC should not go off making half-*ssed concessions on this one detail, > while missing the real points that they need to figure out. Agreed. The question in my mind is how these priority lanes can be used in a competitive context. Can a cable system use them to provide enhanced QOS for their VOD services, while competitors are forced to share the more congested lanes? Can a cable company offer priority lanes to any potential competitor, and if so, should there be price regulation (via Title II) to assure a level playing field? Does the fact that an ISP can charge for priority lanes disadvantage new competitors who may not be able to pay this new toll as easily as an established service like Netflix? Could the priority lane be offered to, and paid for by the ISP subscriber rather than the services that will use it? This assumes that the subscriber would pay for enhanced QOS service, and the ISP could not prioritize bits from ANY OTT service that uses the fast lane. > >> The real problem is the massive investment needed to deploy 20 Mbps or >> better broadband to every home in America. The OTT services have already >> been paying for the WAN bandwidth and CDNs or hosted edge servers. > > Which is fair enough, and I'm not insisting on 20 Mb/s minimum. And by the > way, content to these edge servers, if this content requires massive > bandwidth, can and is delivered "out of band." Which means, the core network > is not always burdened with the traffic volume required to keep these edge > devices well fed. One possibility here is satellite links. I just used 20 Mbps as an example, as it is widely available today, and very useful for moving large files around. We may also consider the reality that in many homes, there are multiple TVs tuned to different cable channels; it is very realistic to believe that a home with 2 kids could be using 3-4 streams at any given point in time. And yes, there are many ways to guarantee QOS to edge servers, which helps make my point that the last mile is the big bottleneck, and will require the largest investment. > > So again to get back to my main point, the cost of deploying useful Internet > TV service should NOT be assumed to be borne entirely by the ISP, as you were > assuming. The ISPs stand to benefit if many of their customers want more > expensive high speed drops, and the ISP can minimize their own costs by > allowing content owners to deploy their own edge devices. Funny how you feel > the urge to debate these points without offering technical details to back up > your disagreements. In the end, only the end user pays. The ISP "invests in infrastructure" and expects to realize a profit from this investment. The same applies to everything in the middle. Whatever Netflix pays for content and carriage, is paid by their subscribers, also with the expectation that Netflix will make a profit. What is left is how the companies that operate the various components of the Internet compete. Some components are highly competitive today, some are oligopolies that try to push costs to the services that must use them, thus increasing costs for competitors. > >> But the backbone is not the problem. It was overbuilt, and is very >> competitive. It's the last mile and the internal routers in the ISPs >> where most congestion occurs. > > Both, Craig. You make this claim without proof. There is intense competition for WAN connectivity, and among CDNs. But this is mostly related to the public Internet. For VPNs, the competitive situation is just as bad as it is with the last mile oligopoly - unfortunately these are mostly the same companies, although the telcos have a significant lead among business customers, while the cable companies are mostly serving consumers. Here are several articles that look at the competitive playing field... http://theweek.com/article/index/257404/why-is-american-internet-so-slow#axzz33xif6Ely http://arstechnica.com/business/2014/04/one-big-reason-we-lack-internet-competition-starting-an-isp-is-really-hard/ http://www.networkworld.com/article/2222196/cisco-subnet/why-does-mpls-cost-so-much-more-than-internet-connectivity-.html http://insights.wired.com/profiles/blogs/new-reports-find-wan-is-failing-business#axzz33xn0FKlI > > Let me repeat an example that you might have missed. ADSL or VDSL, what you > call DSL, can provide (downstream): > > 6.14 Mb/s up to 2.5 miles from CO to home > 12.96 Mb/s up to 0.94 miles from CO to home > > So at least with this telco technology, the main cost of providing a lot more > than just 1.5 Mb/s, to individual homes, IS NOT in the "last mile." It does > not need to be. Verizon might want to deploy FiOS, which certainly is a big > expense in the last mile, but that's THEIR CHOICE. It's not mandatory. I gave up on DSL because it was too slow and congested. Now they are pestering me to come back claiming speeds like those you cite above. Unfortunately these are the theoretical limits. The practical limits are often much less; for example, I live less than 0.94 miles from the CO, but the field tech I talked to told me I would only get 6-7 Mbps. But the real issue is price. I was paying AT&T more than I am paying Cox for 20 Mbps service. The letters I get from AT&T every week offer lower introductory rates, but when the deal is over they are still higher than the cable guys. > > With cable networks, same sort of thing. Delivering high bandwidth to > individual homes can be achieved by assigning more DOCSIS channels to IP > broadband, without NECESSARILY having to make expensive infrastructure > changes in that last mile. If the cable companies get rid, say, of their > analog TV channels, they could buy a lot of last mile bandwidth right there. > It's their decision to make. Also, they could quit wasting bandwidth by > offering both HD and SD versions of the same programs. Their choice, their > tradeoffs to make. Advantage to the Cable companies. As I have pointed out however, the cable companies are investing heavily in reducing the number of homes on each neighborhood loop. They are deploying neighborhood risers all over Gainesville. Recovering analog channels could help, but Cox has too many analog customers to do this for now. Ultimately they will need small neighborhood loops anyway when everything is IPTV on demand. > > On the other hand, what you seem to miss, to feed all those hungry last mile > wide channels will most likely require upgrades upstream. If nothing else, > faster edge servers, which support more simultaneous sessions, as more people > use Internet TV. Yup. But that is what you pay the upstream service for. The OTT services will make these investments to remain competitive, as they have been doing for years. > > So in short, you are overstating this last mile problem, you are > underestimating the core network impacts, and this has nothing to do with > what the FCC should do to ensure net neutrality! The costs of net neutrality > DO NOT NEED to be borne entirely by ISPs, unless they choose to make it that > way. I do not believe I am overstating the last mile problem - this was a consistent theme in most of the links I provided above. Monopolies do not have the incentive to make these investments, especially while they are trying to milk legacy services like "the cable bundle," rather than transitioning to IPTV. I think this response has illustrated that there are costs to upgrade at many levels, not just the ISP. The problem is when monopolies purposely avoid making critical investments, overcharging for an inferior product (both consumers and businesses), and shifting costs to those who have no choice but to work with the monopolists to reach the consumer. Regards Craig ---------------------------------------------------------------------- You can UNSUBSCRIBE from the OpenDTV list in two ways: - Using the UNSUBSCRIBE command in your user configuration settings at FreeLists.org - By sending a message to: opendtv-request@xxxxxxxxxxxxx with the word unsubscribe in the subject line.