KLIA Ekspres train!

I often get my session dropped when using Digi EDGE on that train.

In addition to that, make sure you also destroy every copy of Photoshop in the universe.

Modem is free if you sign a contract before the end of September.

Seems strange in city centre, where almost everyone in their target market is living in high-rises.

The upstream bandwidth gets sold to hosting customers. What's left over goes to consumers.

How do they know where he lives? Just get the modem sent to a friend's house in coverage area, then ring up P1 accounting office and change billing address the first month.

The worst they can do is refuse to listen to your complaining if you are outside their official coverage area. How are they going to fine you?

Where is their tower in city centre area? KL Tower or somewhere else?

The 20GB limit - if they actually implement it - makes me much more likely to subscribe to P1, I think it's a big advantage in this market.

It means that they are going to be able to predict their bandwidth usage within some reasonable range of parameters, and that their bandwidth costs will hopefully be lower than other ISPs'. This means they could provide more reliable and faster service.

Connection gets slowed way down until the next billing cycle. 64kbps or something.

High latency doesn't cause packet loss (unless it approaches what passes for infinity in this milieu).

Different ping programs make their own decisions about when they consider a packet lost.

Standard ping (as found on Unix and Mac systems) will wait forever unless you explicitly give it a limit. This is correct behaviour.

Windows ping - a half-aborted fetus of a program - has an absurdly short timeout, but that doesn't mean you have packet loss, it only means that Windows ping is incorrectly reporting packet loss.


A UDP packet can never truly be written off, in terms of protocol. It may arrive 25 years later. But of course the application will have given up by then.

A TCP packet is considered lost after the retransmit timer has expired. This time varies based on network conditions but I think 3 seconds is a typical starting point.

Which brings us to what I meant by "unless it approaches what passes for infinity in this milieu". When we are talking about a 3-second wait, we're no longer talking about typical broadband latency. Is anyone getting 3000ms RTTs from P1?

Your Astro dish is a receiver only. It gathers the radiation that comes from the sky and lands within a 0.3m2 area and focuses it on a single point - the small receiver unit on the stalk. Any plant with at least 0.3m2 of surface area facing that same direction will collect as much Astro radiation as your dish does, even if there is no dish within 1000km. So your claim of causality is preposterous.

Why would clouds affect it? Unless they are very low clouds...

One thing I find annoying about P1 is that they use some kind of transparent web proxy which doesn't always work.

For instance, right now I can not pull up any web pages using P1, but I can ssh, ping, FTP, everything else.

So I tried the following experiment:

1) ssh to a third host, then ssh to my web server in USA.

2) Run a packet sniffer on the web server, filtering based on my P1 IP.

3) From my PC, telnet to port 80 on the web server. I get a connection, issue a GET request, and then the connection closes. All this time, there is no traffic showing in the packet sniffer.

4) Therefore I can conclude that the connection on port 80 was intercepted and answered by a man in the middle, almost certainly at P1.

I've repeated the experiment with other servers at other data centers and observed identical results. So the problem isn't at the server end. And if I use a VPN or plain proxy to reach those servers on port 80, it works fine.

This is the third time in three days I've seen an outage from the P1 web proxy. Normally they just last for a few minutes but one of them was a bit longer.

It's difficult for me to test, since I only have one P1 modem.

Even though I have turned my modem on and off several times in the 3 days I've had it, my public IP always appears to be 120.140.17.235. Maybe others can chime in about whether they are showing the same IP.


I do have three separate SIP registrations working across the link, which is a good sign. Haven't played with explicit UPnP requests yet, but hope to over the next couple days as I explore further.

I don't think you followed what I wrote.

All requests to port 80 are/were getting intercepted, including to sites like www.google.com.

SSH works fine and always has.

I really don't think there are any "techs" sitting somewhere deciding whether or not to allow each hit to bbcnews.com and yahoo.com.

This post has been edited by rajulkabir: Dec 15 2008, 08:45 PM

Yes, but that's boring. The interesting question is why.

For me it takes under a minute to connect.

It's worth investing some effort in the placement. Both the location and the rotation make a big difference. I found that lying it on its back (with the cables down) always gets me a green signal but it's useless because there's high packet loss then. So the lights alone are not enough to guide you; you need to be pinging or something else to see how well the connection actually works.

I would love to have an audio ping utility that would ping continually and then make different noises when packets are either lost or successfully received. That way I wouldn't have to keep going back and forth to my computer to test new locations.

I am higher than you and I have it working.

But maybe the conditions just happen to be good where I am; I've never tried using it up in any other building.