Technically, adding a router puts in another ‘hop’ and also requires filtering and whatever else the router does, so it does cause a minute increase in latency. If it’s a halfway decent router, this won’t be noticeable. For example, checking my latency with and without router results in a difference of ~0.5 ms. Which is statistically insignificant. That’s like… rounding error or random chance. For pure speed, your router is rated much higher than the connection.
The reason they want to connect directly to your computer for profiling is likely due to wanting to check other things aside from speed and line quality. I don’t know offhand why they would need this info, but to an external system a router can ‘look’ different from a Windows PC. Most routers look like linux or FreeBSD servers. Probably because a large number of them basically are stripped down Linux computers.
If they really need a direct connect for setup(possibly to do setup on the DSL modem, which would require a direct interface), you can just connect for setup and then swap out to your router after setup is finished.
If they’re saying their service will only work if it’s directly connected and you aren’t allowed to use a router or something like that, they’re full of it.
Bytes or bits? I’m guessing that it’s actually the latter. DSL is usually quoted in bits per second (bps) rather than bytes per second (Bps). 3 MBps (24 Mbps) is shit-hot for a DSL link and (hereabouts) requires you to be very near the exchange; not so shit-hot for cable, of course.
The both look like almost full speed if you’re on a 3Mbps link.
It’s either what your paying for or the expected rate at your distance from the exchange.
Its just for testing purposes; they want the router out of the picture to simplify things. They support the line into your house, and may or may not own the modem, but know how to support & analyze it for the most part in any case.
With a router hooked up, for all they know you have 8 computers all doing God knows what with the connection, and any speed test is suspect. Or you’re doing wireless and you have a router that sucks at that or is misconfigured. Or lots of other stupid things that aren’t their fault. One computer hooked up is as simple as it gets and testable, and they don’t have to trust that you have a network set up right rather than tin cans and string.
I keep a spare ethernet cable in my wiring closet where the modem+router is. If I have a problem with the connection, I haul a laptop in there, hook it directly to the modem, and test again (after rebooting the modem and then the laptop), and 95% of the time its still broken, and at that point I call the ISP and tell them this, and it shortens the support call considerably, because I’ve already done every “first step” in their script. At this point they test the connection, look at the modem logs, and all that other stuff they can do on their end. Most times they find the problem very quickly, and I’m back up shortly.
And, yes, I’ve also done this at times and the connection will work fine sans router, which means its not the ISP’s problem, and I have to debug and/or go buy another router.
The reason they do not play nice with routers is to eliminate the potential for misconfigured routers or failures of hardware that they didn’t provide. If you buy a router, the manufacturer or the router should be contacted for support, not the internet provider. Exception being a router combo provided by the internet provider.
Anecdata: I once was doing support for a guy who was complaining of slow speeds, but insisted that it could not possibly be his router, which he flatly refused to remove for troubleshooting purposes. I finally convinced him to do a traceroute; first hop, from computer to router, 3000-some-odd ms. Turns out he had a big shortwave radio transceiver right next to the router, and the little box’s wussy plastic shielding couldn’t stand up the EMF the radio was pumping out.
But yeah, squeegee has it down. A router opens up too many variables for line troubleshooting.
Profiled means that they set you up with a 3MB line profile, and/or performed a study on your line that indicates 3MB ought to be a good speed for you.
Concerning your router, wouldn’t you be embarrassed if you thought you were running off your DSL line but instead had picked up a neighbor’s sketchy wireless signal. Or if something had flaked in the router and it was giving you weird results.
One time I had a router that was so jacked up that I was able to use a packet sniffer to see someone else’s BitTorrent traffic. That is not supposed to happen with DSL, ever, but there it was. Routers shouldn’t have any effect on your speed but you need to remove them for troubleshooting purposes.
I’ll have to ask my son - he set it up. I think he did it your way.
But the AT&T RepairMan was here yesterday.
At my request, he removed the Router and old AT&T DSL modem, and installed a shiny new Motorola modem. Nothing has changed. I just speed tested at AT&T’s preferred site speedtest.net — in NYC —and got the same-old same-old download at 2583 kb/s, which is equal to .32 MB/s.
When you don’t like what you get from one test site, he says, go to another — Speakeasy — (also in NYC), which I did. Here it’s reported in KB and my download was 2078 KB/s or 2.079 MB/s.
I can well appreciate that there can be great disparities in speed testing web sites, but the consistency seems a bit odd. Results yesterday were about the same.
It sounds like you’re having a problem very similar to what I once had. Intermittent loss of sync or very slow speeds. Initially we blamed it on differences in test sites but it turned out that the problem was coming and going while we were coming and going even from one site to the other. I’ve had this problem 3 times, and was told it was a different thing each time, all on their end. One was a bad line card in the remote shelf, once was bad configuration at the broadband gateway, once was bad wiring in the remote shelf. Just keep making the techs work and eventually they’ll figure it out.