Tchibo/O2 selectively throttling download speeds?

inquisitor · 01-06-2012, 13:01

Quote:

Originally Posted by RobF

Or could Tchibo themselves be responsible? Tchibo Mobil completely revamped their customer web interface about 2 weeks ago, and it is still incredibly buggy. Could they also have monkeyed around with broadcasting settings, or told O2 to do so?

It's possible that O2 does apply some sort of throttling but I've neither experienced nor heard from such so far.

Quote:

Added twist: Tried some file downloads using network protocols other than http. In a file DL via bittorrent got a rate of 2 Mbit/s, in two different ftp DL's got 400 kbit/s (Knoppix distro DL) and 150 kbit/s (Index of /), resp. The very same files DL with two different web browsers (Firefox and Chrome, i.e. using http) were again crawling in at a mere 60 kbit/s. So it looks as though a continuous http DL is affected the most in this performance degradation.

These observations rather indicate that there is no intentional throttling since if they throttled protocol-based then bittorent would be the first service to be affected.

Quote:

BTW, can I identify the location of the broadcast tower(s) that serve my cell and get their usage statistics?

You can locate the basestation (or "nodeB" as it is called for UMTS) by finding out its MCC (Germany: 262), MNC (O2: 07), LAC and CellID and then looking it up it one of the many geolocation databases (see mobile - Public Cell ID databases - Stack Overflow).
A pretty straightforward method is to use Google Maps on your smartphone, on which you would disable WiFi and GPS - then localization is performed only by the cellID, which will result in a rather large blue circle shown on the map, which is the estimated coverage area of your serving cell. The tower itself is usually located somewhere close to the edge of this blue circle as it uses sector antennas covering a 120° angle from its location. Once you get a rough idea of where the tower is situated, check the regulatory authority's map for potential tower locations. If requested the regulatory authority does provide the approval documents ("Standortbescheinigung") for the desired location, which reveals the tower configuration (operator, each single antenna including direction, used radio technology and frequency).
But you won't find any load statistics - that's proabably one of operators' greatest secrets.

RobF · 14-06-2012, 14:40

Thanks, inquisitor, for the interesting technical details. I'd also found the location of the 3G towers O2 uses in my cell through this link: o2 GIS BTSMap

It turns out I'm located within 1,500 ft. of 4 clusters of antennae on top of an 8-story building. I can see these antennae from my window - there is no newly introduced obstruction.

I'd also posted the present query in the following forum:
Drosselt O2 selektiv die Geschwindigkeit von html downloads?

The two people who replied so far thought that O2 is selectively applying bandwidth prioritization (through selective throttling & traffic shaping?) prior to exhaustion of the 5 GB allotment of fast access, presently perhaps none to their own O2 Netzclub Customers and plenty to Tchibo customers.

If this problem persists, I'm thinking of jumping from my Tchibo XL plan to the equivalent Lidl plan (prepaid 5GB of HSPA/UMTS access for €15/mo). However, Lidl also uses the O2 network. Are there any reports of Lidl/O2 resorting to the same manipulations as Tchibo/O2 appears to?

FWIW, some more details about this problem:

1. The worst aspect of this is the near 100-fold reduction, since late May, of the sustained speed of html DL of files to some 50-60 kbit/s. Interestingly, this throttling is only applied on a per-thread basis. If I download a file with a multi-thread DL-accelerator, such as the Firefox extension DownThemAll!, each thread will be downloaded at 50 kbit/s, so that if I use the max. setting of 10 threads, the file will come down at 500 kbit/s.

2. A name resolution problem at O2's DNS servers does not appear to be responsible for the slow webpage loading. Linux "dig" records quick lookups (100 msec), and when I replace the default O2 DNS nameserver addresses with those of the OpenDNS nameservers, there is no improvement in webpage loading.

	(#2)
RobF (Offline) Junior Member Amateur Member Posts: 16 Join Date: 20 Oct 2011 Country:	14-06-2012, 14:40 Thanks, inquisitor, for the interesting technical details. I'd also found the location of the 3G towers O2 uses in my cell through this link: o2 GIS BTSMap It turns out I'm located within 1,500 ft. of 4 clusters of antennae on top of an 8-story building. I can see these antennae from my window - there is no newly introduced obstruction. I'd also posted the present query in the following forum: Drosselt O2 selektiv die Geschwindigkeit von html downloads? The two people who replied so far thought that O2 is selectively applying bandwidth prioritization (through selective throttling & traffic shaping?) prior to exhaustion of the 5 GB allotment of fast access, presently perhaps none to their own O2 Netzclub Customers and plenty to Tchibo customers. If this problem persists, I'm thinking of jumping from my Tchibo XL plan to the equivalent Lidl plan (prepaid 5GB of HSPA/UMTS access for €15/mo). However, Lidl also uses the O2 network. Are there any reports of Lidl/O2 resorting to the same manipulations as Tchibo/O2 appears to? FWIW, some more details about this problem: 1. The worst aspect of this is the near 100-fold reduction, since late May, of the sustained speed of html DL of files to some 50-60 kbit/s. Interestingly, this throttling is only applied on a per-thread basis. If I download a file with a multi-thread DL-accelerator, such as the Firefox extension DownThemAll!, each thread will be downloaded at 50 kbit/s, so that if I use the max. setting of 10 threads, the file will come down at 500 kbit/s. 2. A name resolution problem at O2's DNS servers does not appear to be responsible for the slow webpage loading. Linux "dig" records quick lookups (100 msec), and when I replace the default O2 DNS nameserver addresses with those of the OpenDNS nameservers, there is no improvement in webpage loading.