Strange changes of download speed

[AndrewMac]

My machine has Win7 Pro SP1 installed on it, as well as IE 11.

When my cable Internet provider sends a technician to me to check the download speed, he connects his laptop directly to my cable modem with an Ethernet cable and tests the download speed on www.speedtest.net. His results agree with the Internet provider's claims.

Immediately after that he tests the download speed on my desktop (connected in the same way) and receives roughly the same result.

A few minutes after technician leaves, I repeat the test on my desktop and the download speed drops to half of the previous speed. It's inmteresting that such a situation happened to me several times before, on occasion of visits of other technicians.

Summing up:
All the tests were performed with Ethernet cable connected to my cable modem (Hitron):
Test 1: technician's laptop and his Ethernet cable
Test 2: my desktop and my Ethernet cable, directly after Test 1
Test 3: my desktop and my Ethernet cable, a few minutes after Test 2

Essential for the case is to understand the download speed discrepancy between Tests 2&3

I wonder if I can 'inherit' extra open ports or modem settings from the technician's laptop test for a very short period of time, which could justify my perfect download result in test no. 2 and not in the later test.
Please note that download speed isn't essential in this problem

[DaveLembke]

Never heard of an ISP that ramps up the speed for a test on site and then flips the switch to throttle you at a slower connection when they leave. What is the speed your detecting after they leave for download/upload compared to when they are on site.

Are you sure that when you are using speedtest that the location that the loop back is working with is the same between when they are on site and when they are off site?    speedtest usually detects that fastest, but you can select alt sites to loop back against for speed testing.

Are you sure that no one else is competing with you for bandwidth when this test is happening. If you have a player running Netflix or some other computer on downloading or playing web content, or even a Voice over IP phone running it will skew the test results.

Additionally I have tested and detected periods of the day in which the Internet is slower. Latency increases and so due to latency the download/upload speeds are affected. I ran a computer with a batch gathering ping info from my ISP which should be the fastest possible least latency as well as google and yahoo, and then I took the info that was gathered into a CSV file and plugged that data into an excel spreadsheet and graphed it. The graph showed minimum, average, and maximum latency for 1 minute intervals, and you could see a noticeable difference for latency when more people are using the Internet as well as more people are using your ISP backbone. If the test was performed when the Internet and the ISP's network was calm then the speed would be fast, however if an ISP has too many people on each of its network legs, then your forced to compete for bandwidth with others and so you could potentially have less bandwidth than your ISP plan is rated if they have too many people on a leg that your on and they really need to load balance and split everyone up onto 2 or more other legs that lead back to the main communication center.

If you would like a copy of my batch file to gather latency info, i can dig it up when i get home from work to share here. Someone here about 7 or 8 years ago helped me in its creation and its a great tool.

[AndrewMac]

"Never heard of an ISP that ramps up the speed for a test on site and then flips the switch to throttle you at a slower connection when they leave"

Me either and therefore I asked the question "I wonder if I can 'inherit' ....."


"Are you sure that when you are using speed test that the location that the loop back is working with is the same between when they are on site and when they are off site?"

I must say that these tests were done on servers located in a very small geographical area within the same time zone. My ping was always 7-8 ms.


"Are you sure that no one else is competing with you for bandwidth"

I'm absolutely sure!


"Additionally I have tested and detected periods of the day in which the Internet is slower ..."

Surely, this can happen, but we are talking about a 3-5 min period and factor of 4 in speed drop, sort of calamity.

[DaveLembke]

"Never heard of an ISP that ramps up the speed for a test on site and then flips the switch to throttle you at a slower connection when they leave"

Me either and therefore I asked the question "I wonder if I can 'inherit' ....."

Having doubts that they would play this game.

I would test other locations with the same ISP to see if others are also being throttled back as you feel you are. Maybe a friend, family member, or neighbor might be able to assist with this who lives within a few miles of your home. If you see that they have correct speed and you dont then it has to be something at your end. If they too are showing throttled then you have more ammo to go to the ISP with a claim about false advertising in which they are not holding up their end of the agreement for the bandwidth plan that you have with them.

There is nothing you can inherit to give more bandwidth. The control is all at the ISP side by DHCP Lease to modem or a Modem Flash that controls the speed.   

Also I have seen others post similar requests in online forums as a means to try to find ways to unlock faster speeds than that of which your ISP has granted you, hacking the modem etc, and we cant help with this if this is a concealed information request hiding the true illegal intent behind a similar nature legal question. Not stating that this is what you are doing, just stating that we have seen questions that seem innocent that touch on grounds that the information can be used with wrong intent and due to that we might not be able to share specific info that could assist with hacking an ISP in this case.

[Geek-9pm]

On order to make a claim that the  ISP is crating, one would have to get other users to work together to form a complaint.
You will need to ask local people about how to file a complaint.

[AndrewMac]

There is nothing you can inherit to give more bandwidth. The control is all at the ISP side by DHCP Lease to modem or a Modem Flash that controls the speed

In my opinion, we can't say that this is about giving more bandwidth. In fact, the technician results were at the level of my contract and my ISP is technically unable to deliver more bandwidth anyway. It's also worthwhile to note that consecutive tests gave still worse results.

I thought about the inheritance problem, as both of our computers were connected to my cable modem during first two tests and we had to have the same IP address.

I came up with the idea of opening extra ports, as my ISP wanted me to perform several tests having all my ports open and they are usually closed on my PC.

I believe that your last paragraph is completely unnecessary and I have to express my surprise because of it. Your suspicions was a complete miss; I'm not a hacker. Of course, you can find my questions strange or naive but you should know that Internet is to me just a tool to do my job and nothing more.

[Geek-9pm]

For what it is worth.
Some complaints have been made about some companies scrolling the bandwidth of content. For example,  when  ISP the is owned by a movie studio will decrease bandwidth of any download that  contains its movies.

Here is a link to a public complaint made last year.
http://www.consumeraffairs.com/news/att-data-throttling-policy-only-legacy-customers-cause-network-congestion-093014.html

Though AT&T likes to cite network congestion as an excuse for throttling certain of its mobile customers' connection speeds, its latest offer to double the data limits for new subscribers who sign up for its Mobile Share Value plan by October 31 strongly suggests that AT&T's data-throttling activities have less to do with reining in bandwidth hogs than with generating additional revenue for AT&T.

Note that Verizon is also chided.
The above is just one of a number of articles about this topic. 
 

[BC_Programmer]

Cable Internet is typically less predictable in terms of speed because all traffic on a node has a shared bandwidth. What typically happens is that in the evening when people get home from work and kids get home from school, performance will often drop significantly over the speeds you can see during the day. Of course that doesn't explain this behaviour.

One possible explanation could be bandwidth throttling. This is particularly prevalent on Cable internet, partly for the aforementioned reason- if somebody uses up a fair amount of bandwidth within a given period- even well within their contract performance, it will sometimes temporarily "choke" their available bandwidth in preference of other users on that node for a period of time. So the first few tests will go fine, but then it will be choked for an amount of time.