Ditch Your Contract If Your Speed Stinks? > How about enforcing it like the "fleet average"

reply to swhx7

Re: How about enforcing it like the "fleet average"

The new EPA mileage test is supposed to represent typical city or highway driving. When measuring inflation, the test shoppers get a standard basket of goods, designed to represent what most people typically need.

Similarly, the test would sample subscribers who would be verified to have common configurations (invited to volunteer, like with TV ratings) - say, a PC running late-model Windows, with 10/100 ethernet, whatever kind of modem the ISP gives out, and medium distance from the ISP's CO/box/whatever it is.

The PC doesn't make much difference apart from the network card. Things like routes and congestion would average out with different individuals and different times of day and days of the week.

The NIC and the system most definitely make a difference (I don't think my P133 firewall would perform very well if I loaded up Win XP like you suggest). So does the software running on the system and the "integrity" of the software (think compromised system). Unless you want to give these testers baselined test equipment and ensure it's installed correctly, there is no way to accurately test what you are asking to test.

reply to openbox9

reply to openbox9
The simple answer is to run the test from inside of the customers premise equipment or modem. A lot of brands have that capability already.

You mean the test is run by the ISP using tools built into the CPE? If the CPE has the capability, then yes, that's an option. Then you just need to work on the motivation for doing so. And then if you get past that, what's the benefit?

reply to openbox9
OH come on. I work in IT and 100% of our 2k plus machine can do this. It wouldn't be a stretch to say that if your machine is running Windows 2000 you can get the advertised speeds. The ISP have minimum machine requirements, and that should be good enough. As far as other configurations and who does the test, establish as base configuration of 1 computer, within the minimum requirements and go through several different bandwidth testing sites. Take the median values... We can come up with a solution.

reply to openbox9
Yes, and USB cards are pretty slow, too. That's why I'm sure we can determine a good base line machine. A P133 isn't very useful for most people now.

That was my point. Without defining a baseline for a test machine, and ensuring testers have the baselined machine, the test is invalid. And my P133 is chugging along nicely as my BSD firewall with four interfaces. It has no problem passing traffic to max out my 10/1 connection. However, I don't think the same could be said if I loaded Win XP, which was my other point.

reply to jjeffeory
If you work in IT, then you should have a fairly good understanding about how many consumers' computers are comprised (knowingly or otherwise) in the world and how that may invalidate any test that you want to run even if it's on the most robust workstation available today. Relying on ignorant consumers and their minimum spec'd computer (BTW, the specs are for the crappy software that ISPs want you to load and the "CSR zombies", not for connecting to the network) is insufficient for a test that's geared towards providing blame or exposing alleged limitations of ISPs.

reply to jjeffeory
I have 5Mbps cable service and until I tweaked my tcp receive window, I could only benchmark tests at 4.000Mbps. After tweaking my receive window, the speed went up to 4.9Mbps.

I would bet that most customers who have speed beyond 3.0Mbps need to be tweaked before they will get the full benefit of their potential bandwidth.

In my opinion, damn the tests. Damn the regulation. Damn the easy contract out.

What's the best way to improve a company's product?

COMPETITION

The only way we're all going to be happy is if we have choices. Only choice provides a universal motivation for companies to treat customers with respect and provide the best possible product while still making investors happy with a modest profit. Right now most of us don't have a choice. Only the precious few have multiple HSI and video choices. The government should be working with incentives and regulation to foster competition. Oh, and one more thing, we should STOP funding the GD telcos and wireless companies with the USF and make them plow their own fields. If it isn't profitable to be in some rural, swamp town, let the local residents build a muni project or, if they prefer to be like Green Acres where they have to climb a pole for voice, so be it.

reply to openbox9
Lets not forget that a test from the cpe may not be a true indicator of the speed. It was mentioned above that inside wiring is the responsible of the customer, not the telco and there are many types of inside wiring, not necessarily twisted pair, etc. etc....

reply to swhx7

Re: A great sign...

reply to Austinloop

Re: How about enforcing it like the "fleet average"