How are you measuring upload speed?

I have found that some testers seriously underestimate under certain circumstances. The old now defunct BT retail test was the worst, the new one and BY Wholesale test fairly bad. The most reliable for me recently is this site's own: http://www.thinkbroadband.com/speedtest.html
That consistent matches what I get from FileZilla upload to good upload sites (SquirrelSave for example)

When I had server download speed problems because of some bad wiring (down to 12Mbps, briefly 3Mbps), my upload stayed at 17Mbps. However, most testers reported well under 10Mbps.

Some BT users have found that one (or more) resets of the HomeHub can dramatically improve upload speed. Not sure if that is because of establishing a different IP with different routing, or something internal in the HomeHub.

aside: If you unlock an HG612 modem and remove QoS, you may find upload increase from around 17 to around 18.5.

Well I tested with thinkbroad! Originally 17.4 now 7.4.

You need to get an HG612 and unlock it, to give a better idea of what's going on, or just wait patiently until the works on your cabinet are completed.

And it's always nice to have higher figures, but do you actually use 17 upload?

I use cloud back up, but I've never seen the upload speed go above 3, and I have 10 on tap.

Well I mainly upload photos onto Flickr - the upload 2 weeks ago was really fast, but now
the much lower upload speed slows down the process quite a lot. Having said that, all the photos have been uploaded now (33 gig) - only new ones to be taken in the future will suffer.

As for getting an unlocked HG12 - good idea, but I'll be patient & wait.

Ironic really... Having the unlocked modem is almost perfect for situations like this.

The thing that makes it fully perfect is that you have to have it unlocked, and baseline conditions recorded, before things go bad.

When you get really motivated to unlock the modem, it is usually 2 or 3 days too late.

Well I have the stat's already of what my speeds are historically with the Sam Knows box, plus I keep a spreadsheet of my speeds (!).

Anyway even if I have the numbers can I convince OR or Plusnet to do anything about it?

Speeds are an all-or-nothing thing. They prove how fast something can go when everything is working. But when they slow down, they don't help point to *where* they've slowed.

Getting action from the stats depends, as there are a few levels. They certainly can help in getting PN on your side (BaldEagle can attest to that). An Openreach engineer won't make much use of them though - he'll just see the line as it is presented to him at the time of appointment.

If the stats only show you've had FEC/interleaving turned on, then you're not likely to get much joy out of anyone (especially Openreach). However, there's quite a lot of support staff who don't understand the technical details behind this, nor the reason that it is happening more as take-up increases.

In reply to a post by shalom2005:

Well I have the stat's already of what my speeds are historically with the Sam Knows box, plus I keep a spreadsheet of my speeds (!).

Anyway even if I have the numbers can I convince OR or Plusnet to do anything about it?

If it's line related your chances are probably quite high. Where there seems to be a problem is getting someone to look at network issues. Last time IDNet raised my problem with BT they specifically asked for a network investigation and told them the line was fine. The chap who came out knew nothing about network issues and had no information other than a request to fix my line.

Thankfully I wasn't charged even though he found no fault. Dunno if IDNet picked up that tab but I read somewhere that at the moment there is no charge for FTTC call-outs. Anyway the problem is due to strike again in 9 days so I'll give IDNet one more chance then I think I'll switch to AAISP. It really should be possible for someone to resolve this. It strikes every 60 days starting at 1500utc and is clearly network related (fixed by establishing a new PPP session four or five days after it starts).

Edited by Andrue (Mon 10-Jun-13 11:18:04)

Off the subject slightly but:

I noticed a lot of the neighbors have had fibre installed recently and so wanted to check how crosstalk was affecting my line.. I know I was the first one on the cabinet (BT engineer told me I was the only one connected) and I was achieving a full 79999 sync and a achievable of about 97Mb... Now achievable is down to 87Mb and 79889 sync or something weird like that.

The cabinet is ECI and I usually use the supplied ECI modem but swapped to my unlocked Huawei for the stats. Interestingly the attenuation was rock solid at 8.9db instead of 6db. I've swapped back now hoping that it'll return to full sync.

So for me having the unlocked stats allows me conclude that I should try to dissuade my neighbors from ordering/using FTTC because I'm probably suffering from crosstalk.

In reply to a post by timl:

Interestingly the attenuation was rock solid at 8.9db instead of 6db.

I think you mean SNR margin. You should have left the unlocked modem on until it did go back to 6dB imo.