Posts by Phil Thompson 17 publicly visible posts • joined 5 Jun 2007
Many companies have "code powers" to trench in services and erect masts, cabinets etc. Vtesse have trenched fibre from Peterborough to Corby and passed within a mile of me in two directions as they build a fibre network for a bank. Rutland Telecom have a roadside cabin etc in Essendine, Rutland supplying fibre based broadband to a previous slow spot.
Dido Harding of Talk Talk wants to make superfast broadband more expensive. She charges a £10 premium for it when others charge much less (BT / EE / Plusnet to name but three). BT Openreach charge her about £7.50 for superfast on top of the copper line rental she already pays.
She already makes a decent margin on ordinary broadband, but rather than maintaining that is seeking to increase it. See http://goo.gl/SVUzM
BT may be doing a bit of loss leading and sharp pricing, but Harding's Talk Talk and Sky have the biggest margins on FTTC superfast broadband by a country mile.
I don't recall anyone ever saying they were going to provide 100% coverage of 2M broadband, let alone superfast. The UK Broadband wireless "licence holder" guy made that up, interesting negotiating / spoiling tactic. Labour's election manifesto talked of 90% coverage, to put one time marker in the ground.
Lots of BT Broadband users blindly load the bloated CD of "desktop help", "remote support" and other tools onto their PCs - anyone checked the capabilities of that ? Getting it to update itself and phone home with results of a scan for powerline device MAC address or perhaps the client software for managing said devices is also installed and used in some way ?
The ASA *requires* the use of "up to" on fixed speed services such as Virgin's cable services, to indicate the variability due to contention etc.
The ASA further require additional copy for *variable rate* services such as ADSL (Virgin's "national" as opposed to "urban" service) to indicate the variability depending on line length, location, etc..
To compare the performance of fixed rate with a variable rate adaptive service is crass. VM cable modems always connect at the service speed, it's the nature of the technology. Lower throughputs are down to congestion, traffic management and other bottlenecks.
A rate adaptive service on the other hand establishes the best link speed it can, perhaps between 288k and 8M, and then the throughput will be a variable proportion of that link speed. So the "speed" is now characterised by two numbers and the whole advertising department won't understand it. Never mind.
Virgin lie about their ADSL service, saying on their web site that I can get an up to 20M service when no such thing is avilable in my telephone exchange, and further saying that "Unlike some others we're not into capping the speed you get, so we'll always give you the fastest possible broadband we can" when in reality they use an identical BT Wholesale up to 8M service just the same way as every other service provider on my exchange.
Still, it got them publicity which was no doubt the objective.
maybe the median should be used - the speed where half the punters do better and half do worse. That approach works fine for Virgin's *fixed speed* products where you are effectively measuring their congestion levels / performance. The modal speed could be all over the shop from one day to the next.
How to handle rate adaptive products like BT's xDSL ? Any sort of average is fairly useless without reference to the line capability - if the average 'up to 8M' user is getting 4M of throughput that doesn't mean user A on a long line will get more than 0.5 or user B on a short line will get less than 6M. It doesn't tell you anything about an ISPs variability from peak to off-peak.
So perhaps the right approach for xDSL is to report the throughput as a % of the sync speed, to make it comparable with VM fixed speed products. 6M sync, 4M actual, result = 66%. The samknows boxes attempt this by measuring peak speed achieved at off-peak hours.
This is packet splitting and aggregation at the TCP/IP level, presenting multiple parallel connections as a single pipe to client PCs with a single external IP address and combining both upstream and downstream bandwidths to give increased capacity both ways.
A pair of Sharedband Max Premium lines combine to give you a 1.5Mbits/s upstream, handy for remote access or mail servers etc.
Unlike MLPPP it doesn't require the connections to terminate in the same hardware at each end, unlike a Firebrick it uses bog standard hardware costing less than £50 per line at the users end (the Firebrick is also a bit crippled with regards to throughput).
It can easily be used via two or more different ISPs or even different technologies ie a cable line and an ADSL line.
Sharedband's CEO would be the first to say its nothing that hasn't been discussed years ago, but real world practical applications of inverse multiplexing aren't common.
Phil
Using the same bandwidth at the same frequencies under the same regulatory power regime as 802.11 means that WiMax does not really offer an order of magnitude change, only gaining on coding efficiency and QoS.
A short range capability of 40 MBits/s per radio channel (www.wimaxforum.org/technology) divided by N subscribers leads the WiMax Forum's white paper on fixed broadband access to propose "A “best effort” service of 384kbps with 20:1 over-subscription" - sound familiar ?
If BT aren't pushing higher speed and if it is a viable market where are the startups and competing providers ? Why are new housing developments reliant on BT's USO to provide phone lines, rather than wiring in Ethernet sockets and fibre distribution systems from whoever (or cable from Virgin for that matter) ?
You can get pretty well any speed if you are prepared to pay for it, getting it cheaper needs investment and there's no rule that says it has to be BT making it.
Download Speed: 11394 kbps (1424.3 KB/sec transfer rate)
Upload Speed: 636 kbps (79.5 KB/sec transfer rate)
2 * BT lines using Sharedband
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
