Let’s Get These Isp’s To Improve By User Published Uptime Reports.

[ianc66]

Let’s get these ISP’s to improve by user published uptime reports.

A modest proposal lets monitor uptime so we can see which ISP is worth buying – I encourage you to add reports for your ISP – below is how to do and my first report.

It happened again this morning halfway through a post and my CSLoxinfo IPSTAR connection went down I knew it was down because a run a ping in the corner of my screen. Last week it was during an online hotel booking.

I think we all know that the ISPs don't provide the bandwidth claimed

-- Although to be fair it is not solely their problem if xyz.com has a slow server we will get slow response.

It seems to me what is worse is the unexpected downtime IPSTAR here http://csloxinfo.com/index_en.asp claim to have ‘always-on connection’ however as they seem to provide no uptime statistics who knows? Well I want to and I think you do too and I think if together we publish on this forum uptime statistics for any and all of the claimed 24/7 Broadband ISP services we might get some response.

Obviously we can’t do this by running pings and watching 24/7 we need an automated tool and we also to need to monitor more than a ping.

What I have done is downloaded a free server monitor:-

IPcheck server monitor. From here (sorry seems to be Windows only)

http://www.paessler.com/download/ipchecktrial

The monitor runs as a service with its running monitored via a web interface, I’ve installed and set it running to monitor three things

1 My ISP’s primary DNS server on DNS port 53

2 My ISP’s home page monitoring HTTP*

and

3 BBC.CO.UK Monitor HTTP* (you can change this to any reliable out of Thailand server).

*be careful with your choice of web page a large web page can generate a lot of traffic not good if you pay by the MB!

The idea is that if 1 and 2 and 3 fail the ISP’s service is for sure down, if 3 fails probably the link to the outside world is down if 1 or 2 alone fails probably a single server failure. (Strictly speaking we should monitor all their DNS servers but the monitor is limited to 3 sensors) In fact with IPSTAR port 80 fails whilst other things are still running.

To do this accurately you need to run your PC 24/7, make sure your date and time are correct, make sure you discount any downtime caused by power failure or software upgrades to your PC or any downtime caused by equipment not provided by the ISP. I intend to count as downtime any reboot of the satellite modem asked for by the helpdesk (if the service is supposed to be ‘always-on connection’ this is reasonable).

First set of results in the Attachment.

Note that the BBC is on different scale from the other two as I started it later. Note the downtime to my ISPs own server 9.63%

The down time on the after noon of 14 was caused by rain on the 15th note that whilst the DNS was fine HTTP was not.

Enjoy any questions?

ipsaction1.bmp

Edited March 16, 2006 by ianc66

[ianc66]

And two more attachments

ipsaction2.bmp

ipsaction3.bmp

[John_Betong]

And two more attachments

Hi,

No wonder the internet is so slow due to enormous files being unncessarily up/downloaded.

Please take the time to learn about the 'Save As' button and the different file formats.

In a nutshell, bmp files are bit map files which I think means that every pixel is saved individually and has a unique value. Jpg format is a compressed file which I think means that the original picture file is sort of zipped and that similar areas of the picture are grouped together and the similar locations saved as a reference. The algorithm for doing this is called lossy because when a picture is changed the lossy compression algorithm calculates again and the picture size shrinks but the original picture cannot be reproduced. Opening and viewing the file does not activate the algorithm it is only when the picture is altered and saved using a graphics program.

Gif is similar to jpg but does not have as many colour avaiable, 256 is the maximum I think.

Your 1,688 kb bmp file can be saved as 34 kb Gif file using Microsoft's Paint program and selecting 'Save As'.

I have taken the liberty to also crop the unused white space from the bottom of your bmp file, reduced the dimensions from 700x823 pixels, down to 600x434 pixels and saved the file as a Gif. I selected a four colour Gif because it appears only three colours are used. This further reduced the file and here is the 14.48 kb attachement.

As far as the ISP providers are concerned, I think it is like buying a Formula I racing car. The manufacturers test it on a race track, log the speed and sell it. If a person buys it then he is aware that the car is capable of that speed only on a racetrack but if he wants to drive it to work on a Monday morning he does not expect to be able to reach the manufacturer's claimed speed due to the other traffic.

Cheers,

John_Betong

[ianc66]

John,

Thanks for your note I've already been reminded that other files formats are smaller and I'll use that next time.

I'm not altogher sure of your F1 analogy but nevertheless to take it a bit further, if in normal driving a car cannot reach it's full speed potential at least the manufacturer does'nt expect you to have to reboot it's Engine Management System. when the car regularly cuts out.

My particular gripe with IPSTAR goes beyond speed and onto reliability that is what I'm trying to get at. I think reading posts on this forum and others that other customers of other ISPs have the same gripe but for that gripe to be heeded evidence is needed.

When I worked and managed computer systems and networks good managment practice was to test and record performance especially uptime and use that record to fix problems -- not rely on users calling you to say the systems were down. I'm offering a way of peraphs getting the ISP's to start acting as professional businesses and we to be satisfied customers.

Why not try it start recording and posting your ISP's performance?

Edited March 16, 2006 by ianc66

[sierra01]

This is Thailand, do you really think they will take any notice?

[ianc66]

This is Thailand, do you really think they will take any notice?

Well I had my doubts about Thai Institutions ability to adapt but ws heartened by two things. My experience of Dr. Thaweesak and his running of ThaiSARN in 1992-3, when I had at least as good internet experince here as the UK. Second thing many of the ISP are no longer Thai they have been sold to Singapore.

My main doubt was of course that the opportunity would be wasted by faint hearted Farangs.

Load the software and get monitoring your ISP.

[monty]

I think it's certainly a good idea, and doesn't do any harm (ping's hardly eat bandwidth!)

Now I think it's a good idea to specify that it is specifically for iPSTAR users.

With iPSTAR the main gripe seems to be the very high downtime...

With for example ADSL, the main problem or complaint is speed and very seldom downtime. It's working pretty much always, as so far that you have to be unlucky to ever get at your PC while it is down, however quite a lot of times it slows down to a crawl, but without failing alltogether.

I will use that very handy download from you (thanks for the link!), but for a slightly different reason, namely testing the stability of one of my ADSL links, considering that I'm almost 8km away from the phone exchange (nat easy for adsl, especially on Thai phonelines )

[ianc66]

I think it's certainly a good idea, and doesn't do any harm (ping's hardly eat bandwidth!)

Now I think it's a good idea to specify that it is specifically for iPSTAR users.

With iPSTAR the main gripe seems to be the very high downtime...

With for example ADSL, the main problem or complaint is speed and very seldom downtime. It's working pretty much always, as so far that you have to be unlucky to ever get at your PC while it is down, however quite a lot of times it slows down to a crawl, but without failing alltogether.

I will use that very handy download from you (thanks for the link!), but for a slightly different reason, namely testing the stability of one of my ADSL links, considering that I'm almost 8km away from the phone exchange (nat easy for adsl, especially on Thai phonelines )

Monty,

Thanks for your support, to an extent you are right my main interest is IPSTAR but it emphaticaly IS for ADSL also.

I've highlighted the high downtime. Once that is fixed I will turn to performance.

Note the test to HTTP is not a ping it attempts to load the target web page every 15 Minutes if it is slow it reports it as slow if it will not load it reports it has down (I've left the monitors' default times for these measures they seem to correspond to my experience of the performance when it reports a site as slow it seems slow to me). I make the point in my original post to be wary of loading a large page. CSLOXINFO.com is about 59kB.

In fact I am not using the monitor to ping (although that could be a worthwhile test) I am monitoring two web pages and one DNS server, (this is a useful test the packets are small and if your ISPs' DNS servers are down you have a problem)

The bottom line is you can use the monitor to measure poor performance and call your ADSL ISP to account.

Good or bad I urge you to post the results of your monitoring

[sajal]

People use Ipstar only because they have no other alternative(not counting dial-up)....why should someone it better?

[ianc66]

People use Ipstar only because they have no other alternative(not counting dial-up)....why should someone it better?

Gosh I don't know the quiet pride from a job well done peraphs?

The likleyhood that if they don't improve their subsciber base will fail to grow. Is that why they just cut prices I wonder?

The eager Singaporean boss standing behind them?

The alternatives I have today dial-up (very expensive to install a fixed line so far from town) or GPRS not so good either of them.

In the near future WiMax to a fixed line backhaul real competition.

[ianc66]

Well here are the cumulative results since I started monitoring and the last seven days I’m concentrating mostly on downtime.

Bottom line in the last week 3.024% downtime as measured against csloxinfo.com

Comments

BBC performance the monitor seems to be a bit confused and not count some extended downtime – other UK based sites were subjectively better than the BBC when I tried them it is difficult to distinguish between network and web server performance of course, this would be a more useful measure if other people would join in and monitor out of Thailand sites that they regularly use.

Interactions with CSLOXINFO/IPSTAR

1 On Sunday 19 march 16:00 link dropped 2 times seen on ping as destination net unreachable then ping came back but web traffic stalled.

I called helpdesk told them the symptoms explained I could ping their DNS server and the only web traffic I could see was to the satellite modem.

They asked me if I could ping 192.168.5.100 well of course I could as it is several hops closer than their DNS server. Any way it all came back of its own accord. At which point the helpdesk suggested I reboot my computer I replied that I rarely rebooted and thought it rather poor practice, and of course their reasoning was fallacious since I could see the satellite modem web interface the problem was outbound of that point. After the call I realised that they had not asked if my interface to satellite modem was USB (where it might be reasonable to assume the interface could have hung in spite of evidence to the contrary) or Ethernet which would seem more robust, in fact it is Ethernet over a wifi link which runs well.

2 This mailed time and date shown

‘Date: Mon, 20 Mar 2006 11:56:54 +0700

From: iPSTAR Service Center <[email protected]>

To: [email protected]

Reply-to: [email protected]

Subject: Possible Interruption during iPSTAR Gateway Maintenance

For English, please continue downward.

[Thai text deleted for space reasons]

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

Dear Valued Customer,

We would like to inform you that there was unforeseeen circumstance with Gateway Thaicom 4 since 11.50 a.m.

Some iPSTAR users might experience disconnection problem. Please be informed we are restoring this problem

hastenly in order to solve this problem shortly.

For updated information, we will inform you soon.

We are sorry for your inconvenience once again.

Sincerely,

iPSTAR Service Center

CS Loxinfo Public Company Limited

Tel: 0-2263-7171

Fax. 0-2263-8005

E-mail : [email protected]’

Well yes the monitoring did show the problem.

HTTP fails

19/03/2006 15:01:35 6777 csloxinfo.com HTTP HTTP Sensor1: OK 6777 ms

19/03/2006 15:16:31 2464 csloxinfo.com HTTP HTTP Sensor1: OK 2464 ms

19/03/2006 15:31:31 2133 csloxinfo.com HTTP HTTP Sensor1: OK 2133 ms

19/03/2006 15:46:30 1543 csloxinfo.com HTTP HTTP Sensor1: OK 1543 ms

19/03/2006 16:02:31 61776 csloxinfo.com HTTP HTTP Sensor1: Error: Read Timeout (#2) 61776 ms1

19/03/2006 16:03:33 60978 csloxinfo.com HTTP HTTP Sensor1: Error: Read Timeout (#2) 60978 ms

19/03/2006 16:15:17 1856 csloxinfo.com HTTP HTTP Sensor1: OK 1856 ms

19/03/2006 16:30:17 1838 csloxinfo.com HTTP HTTP Sensor1: OK 1838 ms

19/03/2006 16:45:17 1561 csloxinfo.com HTTP HTTP Sensor1: OK 1561 ms

DNS runs

19/03/2006 15:31:55 6 dns server 1 PORT PORT Sensor1: OK 6 ms

19/03/2006 15:46:55 5 dns server 1 PORT PORT Sensor1: OK 5 ms

19/03/2006 16:01:56 4 dns server 1 PORT PORT Sensor1: OK 4 ms

19/03/2006 16:16:56 5 dns server 1 PORT PORT Sensor1: OK 5 ms

19/03/2006 16:31:56 5 dns server 1 PORT PORT Sensor1: OK 5 ms

3 Whilst looking at csloxinfo.com I found this http://csloxinfo.com/contactus/suggestions_en.asp

I will make a couple of suggestions and report the results.

4 Last month I was invited to take part in trials of IPSTAR broadband TV service – I tried it and it didn’t really seem to be of much use to me. However in my reply to the Marketing director I mentioned the poor reliability of IPSTAR and gave him the address of this and another recent thread.

Anybody care to join and monitor IPSTAR more completely or begin monitoring other broadband services?

Edited March 22, 2006 by ianc66

[ianc66]

This is a report of IPSTAR speed from three test I did to the US based Broadband reports.com

http://www.dslreports.com/archive?zip=7711...m=&start=Search

[ianc66]

I'm in an internet cafe.

As of about an hour ago CSLOXINFO had been down since 09:00 Saturday 53 Hours. 'gateway have problem'.

A question is this CSLOXINFO alone -- any other IPSTAR via other providers users out there?

[ianc66]

A report after a further week of monitoring. 22 March 2006 to 29 March 2006

Note that this is IPSTAR 512/256 from CSLOXINFO first installed 17 January 2006 the monitoring methodology is described in the first post in this thread

Well here are the cumulative results since I started monitoring and the last seven days I’m concentrating mostly on downtime.

Note the calculated figures are not right because of 15:20 of down time in monitoring it should read 10 days 20 hours and 12 minutes uptime. For a downtime 24.4%.

And for the last week

Note the program calculated figures are not right because of 15:20 of down time in monitoring it should read 4 days 2 hours and 0 minutes uptime. As against 3 days 6 hours 30 minutes downtime total elapsed time 7 days 8 hours and 30 minutes.

Bottom line in the last week downtime 44.5% as measured against csloxinfo.com

Comments

I had two instances of downtime in my monitoring between 15:05 24/03 and 04:30 25/03 caused by maintenance work on the house, and 06:46 -9:21 29/03 caused by a power failure occasioned by heavy rain and lightening. This is a total of 15:20 in both cases I’ve have calculated assuming IPSTAR was working whilst I was not monitoring it.

Clearly the major item this week was the huge amount of downtime

Interactions with CSLOXINFO/IPSTAR

1 Helpdesk:-

23 March 16:00 web traffic stopped although the ping was OK I rang the helpdesk the support was poor -- problem fixed itself

25 March went down completely just after 9:00 no warning I had checked mail about 8 am --- am on phone 25 minutes till 9:51 at which point down 43 + minutes nothing from helpdesk 10:12 still down no answer from help desk no sms 11:12 still down got through to help desk yes 'gateway has a problem'

I suggested recorded message. Still down but with recorded message at 14:54 still down and recorded message at 17:46

27 (Monday) called helpdesk 13:20 answered still down

28 called helpdesk no recorded message told gateway fixed Monday night had to switch off IPX-5100 disconnect power cord and rx cable now works

Checked mail not a word about the downtime. Also I had signed up for SMS for notification of problems no SMS at any point. The helpdesk told me fixed Monday night however I’m count as not fixed until I had a connection.

2 Last week I wrote ‘Whilst looking at csloxinfo.com I found this http://csloxinfo.com/contactus/suggestions_en.asp

I will make a couple of suggestions and report the results.’

Well I made three suggestions :-

Post an updated version of the printed guide to the IPX-5100 on the web site as it is no longer completely correct.

Post uptime reports on the web site

Post details of the orientation of the satellite dish across the country so users can be better prepared for installation

The response was swift and sensible in the third case my suggestion was not necessary as a site survey would be done by the installer prior to installation (this survey did not happen in my case).

The response was good however nothing seems to have happened.

Anybody care to join and monitor IPSTAR more completely or begin monitoring other broadband services?

[ianc66]

A report, after a further week of monitoring. 29 March 2006 to 5 April 2006

Note that this is IPSTAR 512/256 from CSLOXINFO first installed 17 January 2006 the monitoring methodology is described in the first post in this thread

Well here are the cumulative results since I started monitoring on 14th of March and the last seven days I’m concentrating mostly on downtime.

Note the calculated figures are not right because of 15:20+14:33 = 29:53 or 1 day 5 hours 53 minutes of down time in monitoring it should read 18 days 4 hours and 30 minutes uptime. For a downtime of 19.9%.since I began monitoring 14 March.

Last week results

Note the calculated figures are not right because of 18:27 of down time in monitoring it should read 7 days 5 hours and 19 minutes uptime. As against 0 days 4 hours 8 minutes downtime total elapsed time 7 days 9 hours and 27 minutes.

Bottom line in the last week downtime 2.29 % as measured against csloxinfo.com

Comments

Here there were three instances of rain Thursday, Friday and Saturday 30 March to 1 April, late afternoon and evening whilst the rain and wind were bad on Thursday IPSTAR continued working so good news there. However on the Friday and Saturday local electric power was down for longer than my UPS could support for a total downtime in the monitoring of 18:27 14:33 + the Wednesday 29 March morning downtime already reported for which time I will assume that IPSTAR was running

It was gratifying that IPSTAR continued running in the really bad local weather on the evening of Thursday 30 March, and indeed Friday and Saturday until the power failed.

Overall this week was quite good.

Interactions with CSLOXINFO/IPSTAR

Received email being an Invoice for this month no credit for prior downtime.

Anybody care to join and monitor IPSTAR more completely or begin monitoring other broadband services?

[ianc66]

Rather than post only Wednesday I'm reporting the last seven days of relatively (relatively) good performance -- it is not even three 9s

[ianc66]

A report, after a further week of monitoring. 05 April 2006 to 12 April 2006

Note that this is IPSTAR 512/256 from CSLOXINFO first installed 17 January 2006 the monitoring methodology is described in the first post in this thread

Well here are the cumulative results since I started monitoring on 14th of March and the last seven days I’m concentrating mostly on downtime.

Note the calculated figures are not right because of 1 day 5 hours 53 minutes of down time in monitoring it should read 24 days 12 hours and 26 minutes uptime. For a downtime of 13%.since I began monitoring 14 March.

Last week results first CSLOXINFO then the DNS server

Bottom line in the last week downtime 1.1046 % as measured against csloxinfo.com

Comments

I’ve included the results against the DNS server as it brings out a point it seems that there is some regular shutdown at a little after 03:00 every second day.

At the risk of seeming ungracious whilst this is the best week the results are still rather poor and as those of you who use the service will be aware not complete as they do not reflect some short downtimes.

Interactions with CSLOXINFO/IPSTAR

None.

Anybody care to join and monitor IPSTAR more completely or begin monitoring other broadband services?

[ianc66]

A report, after a further week of monitoring. 12 April 2006 to 19 April 2006

Note that this is IPSTAR 512/256 from CSLOXINFO first installed 17 January 2006 the monitoring methodology is described in the first post in this thread

Well here are the cumulative results since I started monitoring on 14th of March and the last seven days I’m concentrating mostly on downtime.

Note the calculated figures are not right because of 1 day 20 hours 19 minutes of down time in monitoring it should read hours 30 days 10 hours and 30 minutes uptime. For a downtime of 11.08%.since I began monitoring 14 March.

Last week results CSLOXINFO

Note that the figure above is incorrect there was no monitoring for 14 hours 25 minutes, so the corrected downtime is

Bottom line in the last week downtime 1.37 % as measured against csloxinfo.com

Comments

Saturday ( 15 April) saw some very heavy rain and wind and electricity was lost for some time there was a loss of 14 hours 26 minutes of monitoring, interestingly IPSTAR did go down before power was lost presumably the rain was sufficiently heavy to attenuate the signal, as before I have assumed IPSTAR was up whilst I could not monitor.

Interactions with CSLOXINFO/IPSTAR

None.

Anybody care to join and monitor IPSTAR more completely or begin monitoring other broadband services?

[ianc66]

A report, after a further week of monitoring. 19 April 2006 to 26 April 2006

Note that this is IPSTAR 512/256 from CSLOXINFO first installed 17 January 2006 the monitoring methodology is described in the first post in this thread

Well here are the cumulative results since I started monitoring on 14th of March and the last seven days I’m concentrating mostly on downtime.

Note the calculated figures are not right because of 2 day 0 hours 40 minutes of down time in monitoring it should read hours 38 days 10 hours and 10 minutes uptime. For a downtime of 9.1%.since I began monitoring 14 March.

Last week results CSLOXINFO

Note that the figure above is incorrect there was no monitoring for 4 hours 21 minutes, so the corrected uptime is 7 days eight hours and 26minutes

Bottom line in the last week downtime 0.68 % as measured against csloxinfo.com

Comments

Lost electrical power causing stooped monitoring late on the afternoon of the 20th

as before I have assumed IPSTAR was up whilst I could not monitor.

The trend of decreased downtime is pleasing.

Interactions with CSLOXINFO/IPSTAR

None.

Anybody care to join and monitor IPSTAR more completely or begin monitoring other broadband services?

[Joey Boy]

And two more attachments

Hi,

No wonder the internet is so slow due to enormous files being unncessarily up/downloaded.

Please take the time to learn about the 'Save As' button and the different file formats.

In a nutshell, bmp files are bit map files which I think means that every pixel is saved individually and has a unique value. Jpg format is a compressed file which I think means that the original picture file is sort of zipped and that similar areas of the picture are grouped together and the similar locations saved as a reference. The algorithm for doing this is called lossy because when a picture is changed the lossy compression algorithm calculates again and the picture size shrinks but the original picture cannot be reproduced. Opening and viewing the file does not activate the algorithm it is only when the picture is altered and saved using a graphics program.

Gif is similar to jpg but does not have as many colour avaiable, 256 is the maximum I think.

Your 1,688 kb bmp file can be saved as 34 kb Gif file using Microsoft's Paint program and selecting 'Save As'.

I have taken the liberty to also crop the unused white space from the bottom of your bmp file, reduced the dimensions from 700x823 pixels, down to 600x434 pixels and saved the file as a Gif. I selected a four colour Gif because it appears only three colours are used. This further reduced the file and here is the 14.48 kb attachement.

John_Betong

Can you remove that avatar John, its unncessarily clogging up my bandwidth

[ianc66]

A report, after a further week of monitoring. 26 April 2006 to 3 May 2006

Note that this is IPSTAR 512/256 from CSLOXINFO first installed 17 January 2006 the monitoring methodology is described in the first post in this thread

Well here are the cumulative results since I started monitoring on 14th of March and the last seven days. In this post I also show some figures for bandwidth

Note the calculated figures are not right because of 2 day 6 hours 23 minutes of down time in monitoring it should read hours 45 days 4 hours and 50 minutes uptime and 3 days 23 hours 19 Minutes. For a downtime of 8.1%.since I began monitoring 14 March.

Last week results CSLOXINFO

Note that the figure above is incorrect there was no monitoring for 2 hours 30 minutes, and 3hours and 3 minutes of spurious downtime so the corrected uptime is 7 days 4 hours and 22 minutes and downtime 3 hours 7 minutes

Bottom line in the last week downtime 1.77 % as measured against csloxinfo.com

Comments

Most annoyingly a Microsoft automatic update KB900485 caused my machine to reboot at 3am on the 27th (although I had selected Notify me but don’t automatically install), the monitoring software did not restart cleanly and reported 3 hours and 3 minutes of spurious downtime.

Lost electrical power causing stooped monitoring late on the afternoon of the 30th for 2 hours 30minutes,as before I have assumed IPSTAR was up whilst I could not monitor.

The downtime has unfortunately increased after last week’s rather good performance.

Bandwidth

I loaded a test version of this bandwidth monitor and set up a number of bittorrents over last night. Looking at the best aggregate hour 6-6:59 get a download rate of 229,137,944 bytes and upload of 118,768,710 bytes or approximately 497kbps and 257kbps respectively which seems OK.

Interactions with CSLOXINFO/IPSTAR

None.

Anybody care to join and monitor IPSTAR more completely or begin monitoring other broadband services?

[yuyi]

Hi ianc66,

I have just stumbled across your excellent work, really excellent indeed!

Do not worry about the lack of constructive answers, most people here are not able to do anything constructive anyway. (There are some really great guys and gals here though, please don't be offended, you know whom I mean!)

ianc66, I can confirm all your reports, I have seen the same problems.

The last days it was very bad again, most websites need to get pinged before use.

I see 3 separate problems:

1)

Connecting to a site, for whatever action (browsing, email , ftp, ...) does simply not connect. That means they time out "no such site" etc., BUT, after, and only after 10 or 20 pings to that site they start to get through. After that it is ok, at least as long as I let the ping run. However some time without activity (or ping) does let csloxinfo forget the connection again, and we have to restart the ping game.

2)

Then there are dropped packages, 10 - 80 % (yes, that much!), and some days more, some days less. At the moment it is bad, average 10 - 12 %, domestic or international sites alike.

This effect you do not see directly, because when you load a webpage, it does retry, and you just see a slower response.

I'm testing using a simple ping, and without mercy that shows every single package which gets dropped.

These dropped packages do affect domestic and international sites, so it is not just a problem csloxinfo could blame on international connections.

One symptom of that is that often there are corrupted packages arriving, which do for example cut off a ssh connection and show a log entry on the server saying "corrupted package from IP (my.IP) received"

3)

And then you get disconnected again and again, sometimes for a few seconds, sometimes for 15 minutes, many times per day. This is not the satelite link, because usually the nameserver still answers to pings, but nothing else.

----------

For any linux guys out there, I use "fping -elu -p 10000 domain1 domain2 domain3 ..." which gives a nice summary but just a very little load, the "-p 10000" means ping only every 10 seconds, so it is not creating a bandwidth issue, not at all.

Sample result, with real figures from tonight:

------

palmbeachparadise.com : xmt/rcv/%loss = 4717/4411/6%, min/avg/max = 530/853/2309

godaddy.com : xmt/rcv/%loss = 3874/934/75%, min/avg/max = 781/1062/2756

thailand-traveler.com : xmt/rcv/%loss = 4718/4434/6%, min/avg/max = 845/1174/2470

tothailand.com : xmt/rcv/%loss = 3874/2790/27%, min/avg/max = 548/857/2446

eurosport.co.uk : xmt/rcv/%loss = 3874/69/98%, min/avg/max = 964/1204/1563

And yes, the "98%" loss for eurosport is real, page is loading - if it is loading - sloooowwwwlllyyy.

ianc66, or anyone else, please update me here or per PM of what you achieved with csloxinfo, I (and others in my neighborhood seeing the same problems) might join you in your efforts there.

[ianc66]

Yuyi,

Thanks for the kind words.

A few comments:-

No have not got anything from CS Loxinfo I've just learnt to live with it.

I've not posted recently as electric power in my area has been down so much that it would turned into a rant 'Lets get these electric power suppliers.....' Although come to think of it it is the same issue mangement need to think 100% uptime. For the record recent performance :-

Not so good

Some of the specific points you raise I think your point 1 shows poor DNS servers, the delay is getting the address into memeory from disk or have to go to another DNS server I do not use CSLOXINFO's DNS servers.

The cutting of for seconds or minutes yes it happens.

Some of the downtime is the satellite modem for instance the long downtime early on the seventh was fixed by power cycling the modem (and interesting pings work during the downtime the downtime was restyricted to port 80) -- looks like poor code in the modem

Edited June 8, 2006 by ianc66

[astral]

Here is a little chart for you to ponder on.

Any good computer service provider with a 24x7 service should be able to hit 99.99% or even 99.999%

HP has the hardware and software solution.