Android OS Forum banner
1 - 3 of 3 Posts

·
Professor Professorson
Joined
·
214 Posts
Discussion Starter · #1 ·
I wrote parts of this in March 2011 on XDA. Since then, multiple pieces of new information have come to light. Per request of a user here, I am re-posting some of my old info, with some new commentary / thoughts.

Contents:
  • Basic Premise
  • Time Line
  • Proof of Throttling?
  • How would YOU manage a network?
  • Testing
  • Putting it all together

Basic Premise

One of the popular mods for root users is Data Throttle Removal (AKA DTR), as described in this thread:
[MOD] Uncapped Data For Your ROMS (Skyraider, Virtuous, OMGB, ETC). In fact, some people consider it so vital that I have seen them delay upgrading their rom (even if the upgrade contains bugfixes) because they are waiting for a new DTR to be created.

The premise of this mod is simple:
  1. Throttle code was discovered in our services.jar file.
  2. It has been hypothesized that Verizon uses this code to throttle our 3G data speeds.
  3. It is further hypothesized that flashing this mod disables the throttle, thereby returning your 3G speeds to their full potential.

Time Line

The Inc was not the first to get this Mod: it has been released for numerous devices, on numerous providers.

Dec 7, 2010 - [GUIDE][OutDated] How to defeat data throttling... ideas?

Since it has been edited, I don't know when this quote was added to the original post, but:

sino8r said:
SO FAR, THIS HAS ONLY BEEN KNOWN TO WORK ON T-MOBILE USA AND CANADA BELL... ALL OTHER CARRIERS SEEM TO HAVE THESE FILES JUST BECAUSE IT'S A STANDARD IN ANDROID 2.2.1 AND SEEM TO BE CONTROLLED COMPLETELY OTA UNLIKE TMOBILE AND BELL...
ihtfp69 confirms the fact that these are stock Android files, not placed there by any carrier:
ihtfp69 said:
This is code built in by Google. It is not an add on from Verizon. Personally, I would leave it alone.
And in the "How to Defeat..." thread it is stated that the mod no longer even works on T-Mobile: that it is "extinct."

It looks like Dec 26 is when people brought it to the Evo / Sprint world: [MOD] Data Service Governer Killer - Updated with How To

I read through all of that second thread and from the very beginning people were doubting the effectiveness of the mod on the Sprint network, or frankly even the need for it. There are numerous posters in there showing 20-30gb of data usage with no throttle from Sprint even on stock services.jar

All of this should hopefully raise the first question of skepticism:
Would numerous phone providers all rely on the same in-phone files to throttle their networks?

Proof of Throttling?

In the original Inc DTR thread, in December 2010, it was claimed that "Verizon starts to throttle data speeds after 5gb of data. That doesn't mean you get charged more, Verizon just slows you down. This mod will prevent that, and only that."

Looking at the code itself, however, pokes some holes in the original premise:
ihtfp69 said:
Examining the code, if this throttling service was engaged, it would put an icon in the notification bar. You would know it was on. This is code built in by Google. It is not an add on from Verizon. Personally, I would leave it alone.
Looking at your framework-res.apk should show you the icon that is being mentioned. In the stock Froyo framework, for example, it can be found at /res/drawable-hdpi/stat_sys_throttled.png

And here's what it looks like in stock Froyo:


As for the 5gb claims, people were saying this back in December of 2010. There has been no evidence to show that Verizon was engaged in throttling at that point in time. While Verizon had publicly stated that they reserved the right to throttle data (even on unlimited plans), they only stated that it may be done to "the top 5% of Verizon Wireless data users."

Since the time when I originally researched this and conducted tests, Verizon has finally released hard numbers and timetables for their throttling:

http://support.veriz...disclosure.html

When will you begin Network Optimization?
We plan to begin implementing our Network Optimization program in September 2011, though it may take several weeks to fully implement the practice.

How will I know if I'm in the top 5% of data users?
As of August 2011, the top 5% of data users were using 2 GB or more of data each month.

How do I know if I will be impacted by Network Optimization?
A good rule of thumb as of September 2011 is this: If you're on an unlimited data plan, have a 3G device and are using more than 2 GB of data in a month, you're in the top 5% of data users and will be impacted by Network Optimization when you're connected to congested 3G cell sites.

Starting at the end of August 2011, if you are on an unlimited plan, are a high data user and had a contract prior to February 3, 2011, we'll notify you through bill messages and on your My Verizon account if you may be affected.
(Emphasis added)

In other words: between December 2010 and the September 2011, anyone who claimed they were being throttled was incorrect. It simply wasn't happening. Which brings up another important question:

If people weren't actually being throttled during that 9 month time span, how could they have defeated the throttle? You can't cure a problem that doesn't exist.

How would YOU manage a network?

http://support.veriz...disclosure.html

How is this different than throttling?
The difference between our Network Optimization practices and throttling is network intelligence. With throttling, your wireless data speed is reduced for your entire cycle, 100% of the time, no matter where you are. Network Optimization is based on the theory that all customers should have the best network possible, and if you're not causing congestion for others, even if you are using a high amount of data, your connection speed should be as good as possible. So, if you're in the top 5% of data users, your speed is reduced only when you are connected to a congested cell site. Once you are no longer connected to a congested site, your speed will return to normal. This could mean a matter of seconds or hours, depending on your location and time of day.
Notice that the method of throttling that people claimed to have been defeating for the last year (cycle caps) is exactly what Verizon has avoided in their throttling (which wasn't in place at the time these mods were released anyway).

So again, how can the mod actually accomplish what it claims, if it claims to fix a non-existent practice?

Testing

In March 2011, I did my own testing to see if I could find any effect on my speeds from using DTR. Most of what I had read at that time was anecdotal with very few hard numbers. When I did see numbers here and there, they were generally extremely small samples: 1 or 2 speed tests done before and after the mod. Such a small sample is statistically insignificant, especially since we know that 3G speeds fluctuate based on location, time of day, population, etc.

In my own testing, I was not able to find any difference between a stock services.jar file and one that has been modified with DTR. I make this statement based on 70 recorded speed tests, split evenly between stock and DTR. Forty recordings were taken throughout a single morning at my office, where my signal hovers around -84 dBm (the building itself interferes with reception). The thirty remaining recordings were taken throughout a single evening at my house, where my signal hovers around -74 dBm. I flashed back and forth between stock and DTR every few tests to ensure that neither testing condition was clumped into a single time window or boot cycle.

Office:



Home:



If I hadn't run so many trials, I could have easily seen a difference that was purely chance, but falsely attributed to DTR. This is true even across multiple readings. For example, my first few readings at the office:

Stock:
Test #1: 1,155 kbps
Test #2: 1,090 kbps
Test #3: 1,008 kbps

DTR:
Test #1: 557 kbps
Test #2: 1,406 kbps
Test #3: 1,404 kbps

Someone might look at these figures and think that DTR had delivered an increase in max achievable speed. Looking back at the full chart, however, over the course of many trials, it is clear that the differences were natural fluctuations.

Putting it all together

To believe that DTR ever worked on the Inc requires that you believe the following:
  1. Verizon was secretly throttling people for a year, but no one ever broke the story; not even when Sprint came out with their commercial calling out other manufacturers.
  2. Verizon was completely relying on the in-phone throttle files, but was simultaneously manipulating the phone into hiding the throttle icon from the notification bar (since no one has ever reported actually seeing the throttle icon active)
  3. Verizon never wised up to the fact that root users could manipulate these files

Before you even get to testing, these seem like extremely difficult premises to accept. On top of that, if Verizon had any interest in using these throttle files in our phone, why would it take them weeks to roll out their new "network optimization" (AKA throttling) in August/September 2011? And why after a year of these mods being out would Verizon still rely on in-phone files?

Also, consider how the throttling is going to be employed: it will be dependent on your data usage, the cell tower you are connected to, the number of other people on that tower, the time of day, the network traffic, and your contract.

Verizon would not (and could not) count on our phones to reliably measure these variables and adjust speeds accordingly. The type of network shaping that they are describing can only be reliably done from the network side.

Do they rely on our phones to count minutes used?
Do they rely on our phones to count texts sent?
Do they rely on our phones to count data usage?

No. These are all monitored through the network. If they were only monitored on the phone, every hacker and his family would be enjoying free, unlimited minutes, texts, and data.

And yet, we're supposed to believe that Verizon would rely on a few easily hackable files to shape network traffic?
 

·
Professor Professorson
Joined
·
214 Posts
Discussion Starter · #3 ·
Nice. Much improved over the xda version. Let me guess, you just copied from there, and pasted here, and due to the sheer awesomeness of Rootz, it's cleaned up, right? Thanks again for sharing.
Thanks man - my pleasure. And yes, the Rootz software magically edited the content on my behalf; it's so amazing!
 
1 - 3 of 3 Posts
Top