how-to block ads
|
|
Uniqs:
502 |
Share Topic
 |
 |
|
|
| my response. Notice as “Comments (or Reply Comments) – NBP Public Notice #1.”
About the definition of broadband:
1. Form, Characteristics, and Performance Indicators.
a) The form that a definition of broadband should take:
A simple list of set standards based on delivery method. Lets be realistic and acknowledge we have set limits in our infrastructure and cant demand unrealistic stats from a old infrastructure.
b) whether to develop a single definition, or multiple definitions;
Multiple, because we have two very different worlds between wireless delivery and wireless delivery. We can't rule out one or the other to be included in this plan. Also make limits that companies that deliver services via one method, cant begin to deliver them over a different method just to receive funding. Ie: a landline telco invests in satellite because the funding is more available in that market.
c) whether an application-based approach to defining broadband would work, and how such an approach could be expressed in terms of performance indicators.
If the application is 100% third party and not controlled by the isp's. Also keep an eye out for isp's having the ability to detect the testing application, and simply boosting the broadband performance during the application test. This already happens with several speed testing services on the web.
d. the key characteristics and specific performance indicators that should be used to define
broadband;
Latency: speed in ms
Throughput: data in kbps or mbps
Uptime: Time in % per year
Reliability: 100% - ((% of time per hour that Throughput is not meeting minimum specified value. )(% of time per hour that Latency is exceeding maximum specified value.))\2)
Traffic loading: Must be avoided at all costs, A ISP must not interfere with data QOS no more than 5% of packets per each hour. A ISP may only change traffic loading if they can prove that network is operating at more than 95% of capacity, at which time the ISP is given a set amount of time to upgrade their network. Traffic loading should be looked at as an emergency fix until the network is fixed. No single type of data can be sorted differently than any other type.
e. what segment(s) of the network each performance indicator should measure, such as the local access link to the end user, or an end-to-end path;
End to end path, to avoid the ISP attempting to speed up paths that may be known and used for testing. The ISP should be completely unable to detect testing, to avoid false performance data.
f. how factors such as latency, jitter, traffic loading, diurnal patterns, reliability, and mobility should specifically be taken into account;
All should be taken into account on a certain percentage of 100, and each must meet a specified value. Latency, traffic loading, and reliability are the most important.
g. whether different performance indicators or definitions should be developed based on
technological or other distinctions, such as mobility or the provision of the service over a wired or wireless network;
Yes each performance statistic must be developed based on current delivery methods. Although this can hinder the ability of a ISP to offer hybrid solutions to consumers. Also make sure an ISP cant redefine itself to a lower standard, just to lower or avoid operating/upgrade costs.
h. the feasibility and verifiability of measuring different performance indicators.
All data must be measured by a third party. And ISP's must not interfere in the testing ability in any way at all.
2. Thresholds:
a. what minimum thresholds should be assigned to the performance indicators;
Throughput: 3mbps/500kbps for wired services, 1.5mbps/256mbps for wireless/satellite services. This spec is suitable for video streaming, and downloading, any slower speeds severely limit video and high intensive usage.
Latency: under 100ms for wired services and under 500ms for wireless services
Traffic loading: Less than 5% of total packets per hour. Of the 5% of packets that are sorted differently the packets latency must not be any more than 200% of max latency threshold.
Reliability: More than 99% per hour using performance indicator listed above.
b. the minimum thresholds necessary for broad classes of applications to function properly;
Basic web research and browsing, Throughput 768k/128k, Latency 500ms, Traffic loading 98%
Medical monitoring systems, Throughput 768k/128k, Latency 50ms, Traffic loading 99.9%
Low quality video streaming, Throughput 3000k/256k, Latency 500ms, Traffic loading 98%
HD video streaming / large file transfers, Throughput 12000k/1500k, Latency 500ms, Traffic loading 98%
Live gaming, Throughput 768k/128k, Latency 50ms, Traffic loading 99%
Video Conferencing, Throughput 3000k/1500k, Latency 100ms, Traffic loading 99%
VIOP, Throughput 256k/128k, Latency 50ms, Traffic loading 99.9%
Place shifting (ie: sling box type applications) Throughput 768k/1500k, Latency 500ms, Traffic loading 98%
c. whether we should adopt multiple, escalating tiers of minimum thresholds.
multiple.
Tier 1: basic broadband: Throughput 1500k/256k, Latency 500ms, Traffic loading 99%
Tier 2: standard broadband: Throughput 6000k/700k, Latency 100ms, Traffic loading 99%
Tier 3: improved broadband: Throughput 12000k/1500k, Latency 50ms, Traffic loading 99.5%
Tier 4: preferred broadband: Throughput 24000k/3000k, Latency 25ms, Traffic loading 99.9%
3. Updates:
a. what ongoing process should be put in place to update the definition, particularly the threshold levels;
When users use 90% or more of their thresholds for more than 25% of each day.
When new applications can't be adopted by more than 10% of "broadband users" that want the application.
When the "broadband" connection falls below 1% of the users LAN connection speed. Ie: 1gbps LAN = 10mbps broadband. look back to 2004 1% of 100mbps = 1mbps. This has been consistent.
b. how often should such updates should occur;
Every 18 months.
c. what criteria should be used to adjust thresholds over time;
See A.
d. how modifications over time to the definition will affect the Commission’s ability to collect and
publish meaningful data on broadband deployment and adoption.
Broadband has experienced explosive and slow times of growth and adoption, it's a uneven process, that may need to be averaged over many years. New methods of testing may become necessary. | |
| That's waaaay too slow. Basic broadband at 1.5mbps?
Sure that's great for DSL providers, but bad for everyone else. | |
chimera
join:2009-06-09
Washington, DC Reviews:
 ·Comcast
| reply to doublea
Some of those standards make sense, but I still think that people are getting too hung about about latency. Yes, you can define per hop latency within an ISPs network, but it is unfair and unwise to force them to use a third party measure for this since it will inevitably become comprised and meaningless.
I also think that the 1% on LAN speed standard doesn't add up and is just inviting issues. A lot of computers have been coming standard with gigabit ethernet ports for the last five years, but until recently gigabit switches and routers have been out of the price range of most consumers. So how do we define what LAN speed is?
Which brings us to uptime standards. If you're running an organization with critical pieces of medical equipment that require internet access at all times you can't have a single internet connection. Your organization needs a fail over circuit just in case, and that starts to really skew this statistic.
I do agree that we need more transparency, but this shouldn't be built around buzz words like broadband or tiers. It should come from raw data. Yes, ISPs should be forced to disclose information about their average loads, their internal network's latencies any caps or bandwidth shaping systems they have, but we shouldn't go crazy about the semantics. | |
 neowulf
join:2000-10-20
Port Orange, FL | I do agree that ISPs need to be more transparent. It is hard to define something that will become policy when so much is unknown.
Pretty much I have a feeling the definition for "broadband" is still going to be considered much lower in standards then we would like to see. I mean it is 2009 and some how I feel we are going to see what some of us in 1999 didn't consider acceptable.
My biggest concern is that some how they are going to define it so satellite providers are still considered broadband. Maybe I am bias for having to have been on satellite "broadband" before, but there is no way anything about satellite internet fits the definition.
You might have mentioned people getting too hung up about latency, but if there is no requirement at all in that respect it opens the door to companies that have no business being called broadband providers access to federal broadband dollars.
Latency is also very hard to explain to people, and I have a feeling that satellite providers would use the argument, 2000ms is only 2 secs. Do you think a extra two seconds to access your email is really that bad? | |
chimera
join:2009-06-09
Washington, DC Reviews:
·Comcast
| I'm of the opinion that wired and wireless broadband should be held to different standards which is a nice way to prevent satellite providers from arguing that the entire world is properly served by our current setup.
As for the issue of latency, this is a tricky issue and to be honest it's tricky enough to make it so I don't want it included as part of any standard. Often times users see high latency when there are routing issues on a network which is a sign of failing equipment or bad design, but in other situations it's based on how the technology works or geographic limitations. When your latency is extremely high it is still possible to perform most business functions provided your packet loss remains low, and when we get down to it that's more important than gaming (yes I game).
The only common businesses applications that get hit by high latency is VoIP (which requires low latency for a proper chat) and web browsing, which is only effected by extremely high latency. Personally I wouldn't have a problem hosting a file server for hundred plus megabyte files on a connection that had enough bandwidth and low packet loss even if it had 10,000ms latency.
The real trick with latency that you have to think: What takes more time: getting your request there and back or having the content delivered.
For a site like Google.com that's 16KB it can be downloaded in 128ms on a 1Mbps connection so latency is your limiting factor. When downloading a 1GB movie from amazon bandwidth is your limiting factor so a little extra latency isn't that bad. | | |
|
1 edit | Open, untainted, everyone should get at least 1Mbps/200Kbps. unless its illegal activity (ie: sale of drugs, music server allowing illegal access of, say, music, viral site); everything on the net should not be blocked or slowed in any way and this should be a condition of being a provider. Everyone should get at least 1Mbps/200Kbps. | |
|
