Well hello, I have been looking around this forum for quite a while now (along with several DSL site's) trying to find a means of assessing, and if need be tweaking my 256 kbs ADSL connection, and after using information gained and ping plotter (v2.41), over a 12 hour period I seem to have some interesting results, but I'm not quite sure what to make of them, could I possibly PM one of the administrators attaching the results and any further details required so that any suggestions, comments or tweaks may be posted to help me in my quest?

Many thanks in advance!

P.S. I also think that adding the following information as an "ADMIN" post to your forum would be of help to other future posters, as without doing this I was unable to make this post from behind Norton F/W 2003! (found on another forum that I visit, but URL's changed for here).


Important! Users with Norton Personal Firewall
-----------------------------------------------------------
If you have your firewall on, you will not be able to post. Here is the solution to fix this.

Windows XP users with "Norton Personal Firewall" software, or similar (adding a trusted site):


Open NIS.

Click Options.

Click Internet Security (or Personal Firewall).

Click Advanced Options.

Click the Web tab.

Click Add Site. A new site/domain box appears.

Enter the url of the site that you want to receive the referrer information and click OK. That site name appears in the left frame of the Advanced Options window.

Click the name of the new site.

Click the Privacy tab.

Check the "Use these rules for..." box.

Change the Referrer from "Block" to "Allow."

Click Apply, and then click OK.

Click OK to close the Options window.


Also If you wish to disable all filtering: Click on custom settings and make sure enable privacy is checked. Then click on the Custom Level tab. Under Confidential settings select medium. Under Cookie Blocking select None:Allow Cookies. Do not place a check mark next to Enable Browser Privacy. Place a check mark next to Enable Secure Connections (https). Click OK and you're all set.

In Norton '03 the steps are slightly diffrent.

1. Open the NPF or NIS main window and click options> Internet Security

2. Move to the 'Web Content' tab

3. Click the 'Add Site' button and type 'www.pingplotter.com' w/o the quotes. Click OK.

4. In the site list to the left click on the new www.pingplotter.com entry.

5. In the 'Information about visited sites' section on the right clear the 'Use Default settings' check box and click the 'permit' radio button.

6. Hit ok and close the main window.

NOTE: You may have to wait a few moments or close and restart the browser for the change to take effect.

Also you must know the Supervisor password if one is set.

-----------------------------------------------------------

Thanks for the referrer check instructions. You're the first person that's asked / commented about this - so I just shut off our referrer checking until the time (if) some incident happens that encourages us to turn it back on again.

If you send your results to support@pingplotter.com, we'll be happy to check them out and comment on them. If you'd like, we can post the answer back here (along with some screenshots), or we can keep the conversation offline - just let us know what you'd prefer.

- Pete

Hi Pete
Thanks for the speedy return, I shall e-mail the results later tonight/ tomorrow (going out for a beer now), it would be ok to post the results as screen shots and any suggestions you make (may be of interest to other users).

As for your referrer checking, if it has done the job that you required of it in the past, turn it back on and just make a post that will stay at the top of the list to let people know, as other forums use this system as well, so it must be of some use!

<img src="/forums/images/icons/cool.gif" alt="" />

(the following was sent via email by the original poster)

The problem is that my ISP here in the UK (Tiscali), is always being complained about in several of the forums I have visited for its service and speed etc, I must admit the service is not too bad the tech department are quite reasonable for a call centre in India, but the speed is up and down, to which I assumed was due to the volume of “traffic” on the web, and the connection times can vary from 5 min to as bad as 3 hour’s which for an “always on connection” is quite poor and annoying, so I decided to do my own research and tests which I could eventually pass on to the Tiscali tech team and try to sort something out, but to do this I must first be able to fully understand the results my self and try every thing possible to improve performance at my end of the line.

( ... many hardware specifications snipped for the time being as they don't have a big affect on PingPlotter results ... ) ... the modem is a USB Sagem [email]F@st[/email] 800, all drivers are up to date.

The test using Ping Plotter was carried out on 11/12/2003 between 02:09 am and 14:09, no uploads or downloads were made during this period, (due to being asleep), connection to the internet is via DUN

On reading the results the problem seems to be in between my PC and the first “hop”, but as to the cause, it may be my settings, my modem, the filters, the extension, a line fault, I’ve no idea, but any suggestions you make will be welcome.

( note from Pete: I resized this image and added the blue ellipse to highlight the area discussed in the response in the next post).

The data you sent me actually looks *quite* good. The hop 1 data does have a pretty wide range of latencies, but that's often normal for hop 1. Hop 2 and 3 look solid and consistent (except for one spike at 1:20ish PM - and an isolated spike often doesn't indicate a problem).

You always want to start at the final destination and look for periods of problems. If there is any problem at the final destination, then you look back at upstream hops for the same problem shown in those hops. The first hop that shows the same problem is likely close to the source of the problem. The 1 spike you had at 1:20pm originated at hop 1 (that's the first hop that shows the same spike), so it's pretty likely that something between your computer and that hop is the cause of that spike. This spike is only one sample in length, though, and isn't severe, so this in itself isn't any reason for concern.

My in-brief analysis of the data is that there isn't a latency or packet loss problem over the period of time you collected data. Were you experiencing any poor application performance during this period?

Based on this data, I would suggest that you ignore hop 1 in your analysis unless there is a problem in a downstream hop. You also should correlate other problems with this data. If you're playing a game and getting poor performance in-game, then look at the PingPlotter data you're collecting concurrently to see if PingPlotter shows any numbers that back up your feeling of poor game performance. Maybe some other application is performing poorly (email, http, ftp, etc), or your network just feels generally sluggish - correlate that with PingPlotter latency and packet loss data.

Once you've found this period where you've correlated symptoms with numbers in PingPlotter, then it's time to look at PingPlotter to find the likely culprit. PingPlotter might have a certain "signature" that you can recognize (packet loss / latency spikes - I use the term "signature" somewhat loosly here, just as something that you can recognize when you see it). Once you identify that signature, then you might not need to correlate PingPlotter numbers with real-world symptoms, and you can just run PingPlotter 24X7 to look for the signature. You can then troubleshoot with PingPlotter looking for that signature, and trying to determine the cause of the problem.

The data you sent me doesn't have either side of the equation here, however. It doesn't have the symptoms (ie: description of sluggishness or similar application problem) or any "signature" data that I can recognize. In general, I'd say that your PingPlotter data looks quite good (if you ignore hop 1, which you *should* do in the case of the data you sent me, I think).

I would suggest as the next step that you collect data continuously - and look for a specific performance problem that feels like packet loss or latency - and then look to PingPlotter for the hard numbers that will back up your claims so you can bring that to your ISP.

Feel free to respond here or to the support address with any ongoing questions or any followup data that you'd like to add to the story here.

Hi thank's for you opinion, I think I'll start first by running a plot over 48hrs while browsing, to see if there is any "signature" that may link with any performance problems. After that I will start to look at running a test direct from the master phone socket, having disconected every thing else from the line (eliminating the extension, microfilters and any interference from my Sky Digital box), so it would be possible to make a comparison.

<img src="/forums/images/graemlins/smile.gif" alt="" />