Thanks for bearing with me on this - when I say "the ISP's figures" I mean figures quoted by them in emails to me. There isn't an online status page (but after this I am certainly considering moving to a supplier that does provide one).

The "router figures" are the ones on the Show Statistics page generated by my router, and the "Netmeter figures" are the ones compiled by Netmeter.

Have you tried the download test yet?

Yes - and the UL and DL rates which NetMeter showed while the test was in progress fitted with the ones reported at the end of the test.

This, and the figures from the router, are making me think Netmeter is quite accurate. I'm going to have to go back to the ISP and tell them I think their figures are out.

Include the Netmeter stats from as many days back as you can which may help them identify if and when there was some miscounting at their end.

Quote
There isn't an online status page (but after this I am certainly considering moving to a supplier that does provide one).
Unquote

Ask them to send you a graph of the usage, they will have this. Emails are sent when approaching and when reached, and in their portal they should be able to see the traffic on each given week.

Hello again

After 2 1/2 months Telecomplete have come back on this, saying

In reply to:

---

The monitoring we perform is more accurate than what the EU (end user/customer) is performing; we can monitor the entire line on the hour every hour instantly. The EU's checks aren

I don't think you would be too far out to assume 5-10% network overheads.

I assume you are only interested in traffic down your ADSL line. Your router/modem is probably as accurate as any system your ISP uses.

Perhaps you should change to an ISP which lets you monitor your usage in (near) real-time.

Dugeen
Earlier you said "the ratio of UL:DL packets reported by the router is about 1:1.3". This is comparable to the 1:1.13 for WAN packets on my Netgear DG834GT router.

There is no way this can represent the true upload/download ratio for bytes (as distinct from packets), so it seems to prove that not all packets can be the full 1458 (or 1500 or whatever) bytes long. In other words, light users like you and me are sending/receiving a lot of small packets. Heavy users downloading large files should fill their download packets with the full 1458 bytes per packet.

The point of this maths is the thought that maybe Telecomplete are calculating your usage by counting packets and multiplying by 1458 to get bytes. And therefore getting the wrong answer.

Just speculation.
Ian

{Edit} Or it could be the other way round - all short packets could be padded by the router to make full 1458 byte packets. In which case Netmeter will be underrecording the number of bytes sent/received and Telecomplete will be right.

Someone here will know which way round it really is.

Edited by deleted (Sun 14-Oct-07 23:31:34)

Also could be worse if they have an odd MTU value set.

AFAIK most ISPs do not do a simple packet count, most are within a tolerable range of the end-users own monitoring.

In reply to:

---

The point of this maths is the thought that maybe Telecomplete are calculating your usage by counting packets and multiplying by 1458 to get bytes. And therefore getting the wrong answer

---

That is the exact possibility I thought of, and I've been regularly calculating a 'padded total' using that method. It averages about 2x the amounts logged by the router, but the amounts quoted by Telecomplete are still higher than that.