ISP rant

[jay]

Mark,
that is -L 5000 but the L has to be lower case and no space between the L and the 5000 so
-l5000 which makes it look like 15000

Interesting enough 1500 however is the MTU size of Microsofts ethernet interface and the maximum size of 1 ethernet packet is 1514.

for the heck of it - I tried it but got an error in the [-15000] entry.
What did I do wrong?
Should I have entered [-1500] instead?
Click on screenshot to enlarge if needed>>>

[poppo]

Well guys, I hate to say it, but you may be right about the throttling. I know these folks are low lifes (i.e. wanting me to pay for replacing a cable they screwed up), but I really didn't think they would go to such extremes of blatant lying on such a wide scale.

What caused me to possibly change my mind was after looking over their site, I found a new 'service'. All of their packages always stated the all to common 'up to' xMp speeds. But now you get to pay more to get what you are already paying for.

The Avalancheâ„¢ Power Boost  - $10.00/mo.
Choose the Avalancheâ„¢ upgrade to give each of the above services an extra boost of downloading power. Great for movie downloading or large file transfers!

So it looks like you get to pay more not to be throttled (as much).   Any of the higher packages should streem or download just fine if you were getting the advertised 'up to' speeds.e
I'll still wait to see what replies I get to all of the screen shots I was asked to send, and my direct question about the throttling. I'm expecting I either won't get a reply because they know they have been caught, or maybe things will mysteriously fix themselves (they unthrottle me). That's how I originally got the higher tiered plan for the cost of the lower one (by catching them lying about things during my pre-sale questioning).

I suppose the main reason I was not believing it was throttling was because of the extreme amount of it. As I noted, properly implemented throttling should not negatively affect general web browsing if you are not doing anything else.

And I've seen their 'headquarters'. I really don't think they have a lot of sophisticated monitoring. From past experience it seems that they only respond to problems after enough people call in to complain. I'm sure they monitor for total outages, but I'm not sure they monitor for bottleneck issues on a single access point.

[poppo]

that is -L 5000 but the L has to be lower case and no space between the L and the 5000 so
-l5000 which makes it look like 15000

Hmmm.... it's been a while since I used ping options. But on my XP machine it doesn't work unless you leave a space between the -l and number.
Ping -l500 does not work. Ping -l 500 does (and correctly shows a size of 500 bytes)

[StatFreak]

I cannot send 5000 bytes to Google without it timing out (note the 10 second delay and the maxed out TTL.) I can't even send 2000. However, I can send 5000 to NLG and get a response. I basically have DSL speed (about 768kb-1Mb download speed)

<ADD> You can send 5k bytes to NLG and compare your response times to mine, if that would help. The larger packet size will magnify small delay differences and should be more useful.

You might also try running some tracerts to see if there are timeout problems and if so, where they are located.

[brichter]

Hmmm.... it's been a while since I used ping options. But on my XP machine it doesn't work unless you leave a space between the -l and number.
Ping -l500 does not work. Ping -l 500 does (and correctly shows a size of 500 bytes)

This difference is probably between an upgraded version of WinXP and a fresh installation, the upgrade doesn't change some components. My personal peeve is the command prompt, with the older version you simply click and drag the cursor to highlight text so you can copy it, with the new version you have to right-click and select "Mark" before you can highlight the text. Why the he!! did they add an extra step?

[StatFreak]

Hmmm.... it's been a while since I used ping options. But on my XP machine it doesn't work unless you leave a space between the -l and number.
Ping -l500 does not work. Ping -l 500 does (and correctly shows a size of 500 bytes)

This difference is probably between an upgraded version of WinXP and a fresh installation, the upgrade doesn't change some components. My personal peeve is the command prompt, with the older version you simply click and drag the cursor to highlight text so you can copy it, with the new version you have to right-click and select "Mark" before you can highlight the text. Why the he!! did they add an extra step?

Yes, that sucks, but for keyboard oriented folks like me, it's much faster to hit  <Alt><Space> , E  , <Enter> .  Then highlight the text (it highlights in the old block format, not as lines of text, via the standard method of using the cursor keys in combination with the <ctrl>, <Shift>, <Home>, <End>, <PgUp>, and <PgDn> keys.) and press <Enter> once more to copy the data to the clipboard.

To paste text from the clipboard, press <Alt><Space> , E , P


Both of these are much faster than using the mouse, once you get used to it, since you should be typing on the keyboard with both hands and not holding the mouse at all when in a command window.    Of course, for those one-fingered typists who never take their other hand off of the mouse, use the right click menu.


Stat

[brichter]

I cannot send 5000 bytes to Google without it timing out (note the 10 second delay and the maxed out TTL.) I can't even send 2000. However, I can send 5000 to NLG and get a response. I basically have DSL speed (about 768kb-1Mb download speed)

<ADD> You can send 5k bytes to NLG and compare your response times to mine, if that would help. The larger packet size will magnify small delay differences and should be more useful.

You might also try running some tracerts to see if there are timeout problems and if so, where they are located.

What? They're actually blocking ping floods somewhere on your ISP's network? How dare they!  
 

This is a holdover from days gone by, when oversize ping packets were used maliciously to hammer older devices by forcing them to fragment and reassemble the packets. There's a fair bit of overhead involved in calculating/writing the fragment offset header on the TX side and reassembling on the RX side, and there was the old Ping of Death" bug that would overflow the RX buffer of older OSes by sending a packet larger than 65535 bits, which is the maximum IPv4 packet size. Malicious ping floods can still use large data payloads to DoS slower computers/broadband connections, as even though it's only a ping request the receiving side must process the whole request (including all the data contained within) before it can respond to the request. With the processing power of modern computers this isn't an issue any more, the computers we have barely bump the processor even when processing traffic at the highest broadband speeds available.

The max Ethernet frame size that Jay referred to above (1514) is correct, usually the MTU (Maximum Transmission Unit) used by operating systems is 1500 bytes. This allows room for the IP headers to be added without forcing fragmentation of the packets. In some cases (sending traffic over a VPN for example), it is necessary to lower the MTU to allow for the extra headers added as a function of encrypting the traffic.  

[brichter]

Both of these are much faster than using the mouse, once you get used to it, since you should be typing on the keyboard with both hands and not holding the mouse at all when in a command window.   Of course, for those one-fingered typists who never take their other hand off of the mouse, use the right click menu.


Stat

It depends on how you're working. If you're inside a file, your method might be faster depending on where the cursor is and where it needs to be.You still need to use the mouse to get the cursor to your starting point if it's not on your current line when at the DOS prompt in a command window, and I can do that faster with the mouse. I've had several developers at work lose this challenge to me, it's always good to get a free Thai lunch now and then. Mmmmm, Thai...

For editing text files, I use vi. None of these crappy Winblows finger contortionist key combinations for me.   I wouldn't use Windows at all except some of our products and UI interfaces require it. Damn you, Bill Gates!

[StatFreak]

...where the cursor is and where it needs to be.You still need to use the mouse to get the cursor to your starting point if it's not on your current line when at the DOS prompt in a command window, ...

Actually, you don't, but you're right that it depends on program you're in and your location in the text as to which is faster.


I admit that I tend to use the <ctrl><shift> E,P  to paste paths and domain names into the DOS window more often than I use the mark function to copy text out of it.

[brichter]

In my case, I'm usually copying text out to show developers how bad their code really is... 

[jay]

Just coming back to topic, why not spend the $10 for the speed boost. Then you will know for sure. SInce there is no contract you should be able to cancel it next month if it doesn't change things.

$10 is much cheaper than high blood pressure pills.

[poppo]

Update:

Surprisingly I got a VM (I was in the shower) and they had me change my IP address do to a 'possible conflict' with someone else on the network. Now, I had actually asked them that once before when I was having some dropout issues on the old tower and they 'assured' me it was not possible. Well, anything is possible, but you would think their system would alert them to IP conflicts.

What I suspect (since they did not answer the question about throttling), is that maybe they moved me to a block of addresses that are not throttled. I'll find out this evening if the 'fiix' worked, regardless of what it was.

[poppo]

Well, the IP change didn't fix anything.    I still think they have something screwed up. When things slow down, I am seeing ping times to the gateway (on the tower) go up. Generally that is an indication of a misaligned antenna. But it ONLY happens during prime time and it's not constant. Normal pings are 5ms-11ms all night and all day. But now they will fluctuate from 9ms to 80ms during peak hours. It never used to do that even when it was slow in the evenings. Maybe it is throttling, but it just seems like the link can't handle the load or the traffic is not being prioritized right.

[StatFreak]

Maybe "Mom" and "Pop" need to hire a professional and find another job for "Junior". lolol         

[poppo]

Totally down now. Tech support lines are 'all circuits are busy'. I am doing this on my iPad.

[poppo]

In a slight twist of events, after the system came back up last night, speeds were excellent for about a half hour (this was during the normal slowdown period). But then they started to slow to a crawl again (I'm assuming as more people got back on line). So that pretty much blows their 'atmospheric' theory out of the water. I sent yet another e-mail explaining what had happened and how a 'traffic issue' could really be the only explanation. I actually got a reply this time stating that they are working on the speed issue (must be others complaining too) and they are going to add 2 more access points on the tower in the 'very near future'.

So I'll wait a couple of weeks and see what happens.