Has anyone else noticed that rootmetrics usage doesn't count against your allowance?

Hmmmm, I wonder why?
Hahahahaha

Money hatting

Well that kind of confirms what my brother in law who works for Voda network operations team told me months ago.

That EE pay huge amounts of money for the fully detailed reports which is all the tests that Root metric do along with all the users apps tests, This includes lots of info like gps accurate location,cell id and surrounding ones,signal levels and of course the speed tests.

Vodafone have just launched their own app to do the same thing and more as runs small background tests and logs all cell ids. This enables them to see where demand is and also helps correct the coverage maps especially with 4G as can see if the data sent back from handsets matches up with the computer models. The good news is that the app works very well and uses hardly any battery and network usage is reasonable along with not counting towards usage limits.

The more people that install the app the better feedback that they can get about the network. At present i believe it is only available on IOS and Android.

What's the name of this app please? I'd like to feed my their data speeds in Ashbourne, Derbys. so they can at least upgrade the town to 3G. GPRS is shocking!

It still can't produce magic for EE though, what you see in the results is fact. Somebody posted some right tosh straight from Vodafone's PR page before stating it was fact that they had the best network, and calling into question Root Metrics, however as I pointed out the press release by Vodafone only made very limited claims and was carefully worded indeed.

Recently Ofcom has conducted an extremely technical analysis into coverage and found Vodafone the only network to be in breach of Ofcom coverage obligations. Other independent tests by Broadbandgenie showed similar results. Just because EE commission some of these reports doesn't mean they are false or inaccurate. The devices they use are off the shelf, the tests are the same and the areas they test are defined by the Eurostat map zones.

Hmm

If paid some money would get at top, then on the opposite...

How much money did Vodafone ask from RootMetrics and Ofcom?

Really? Because somehow the Rootmetrics app also consistently gets much higher speeds than other speedtesting apps, suggesting some form of extreme coincidence or traffic prioritization.

Given the lack of charging indicates they are clearly identifying the traffic and treating it differently from normal traffic, giving it a little "boost" isn't as far fetched as implying they treat it entirely fairly in all other respects.

I presume you've got a few thousand results to back that up?

The lack of charging does not mean they are clearly identifying the traffic at the time in any way. Most Zero Rating/Charging is done by the charging systems based on the destination addresses, rather than in real time.

The Vodafone app is called NetPerform and tells you also about data usage from each app.

With regards to the Root Metrics app I have also found that the speed tests on Vodafone seem to be higher than on other tests.

Although I dont expect Root Metrics to change the results of EE connections what is obvious is that by EE buying the data is what areas they do their testing in hence why some places are now getting done for the 2nd time in a year yet other places have not been done at all like Northampton,Norwich,Peterborough,Aberdeen,Plymouth,Bournemouth,Cambridge. All are large cities with high amounts of mobile users but none of them have EE 4G

No, I don't have thousands, I have several hundred though, which if you know anything about stats is plenty (in this situation) to deem a statistically significant result.

As for charging - no it doesn't have to be in realtime, but we also know they have realtime control linked to charging systems - if you run out of allowance, certain destinations are instantly blocked while others are still allowed.

Fact is
a) They have the appropriate network equipment in place to prioritize or prioritize traffic in real time, according to type or destination and are using it (e.g. P2P)
b) They have the appropriate rules in place to identify Rootmetrics traffic whether or not in real time and are using it
c) They have the appropriate equipment and rules in place to implement real time traffic controls (i.e. throttles and blocks) according to information from the billing system and are using it.

Doesn't take a genius to imagine they might be using all 3 capabilities co-operatively.

A) Yes-ish. It's clear they have appropriate mechanisms to throttle traffic, not the same as priority in any network sense.
B) Yes, but if it's not a real time capability it can't be used to "boost" the performance.
C) throttling/blocking yes. Still not the same as prioritising traffic in real time (end to end) through the network.

All this is still ZERO evidence that they can and do prioritise speedtest traffic in real time.

You've also presented ZERO evidence that they don't.

All I've said is there's evidence the Rootmetrics app gives faster results in a consistent manner.

There is also evidence they have the capability to treat Rootmetrics traffic differently.

That suggests prioritization is a possibility.

As for A) and C), throttling (deprioritisation) uses the same mechanisms as prioritization. From a network administration point of view it's configured in the same way, usually even on the same screen/box/application/script. The ability to do one automatically infers the ability to do the other. The network does not care if you set one type of traffic to priority 3 and "all others" to 2, or whether you set one type of traffic to 2 and "all others" to 3. The capability to do so is one and the same.

I have indeed presented ZERO evidence that they aren't. Then again I'm not trying to make that point, simply that your point that they they are identifying it and treating it differently also has no evidence. (All operators zero charge for lots of reasons, e.g portal/bundled streaming etc, are these all also implicitly boosted?)

Throttling and prioritisation are usually done entirely independently by different mechanisms. Prioritisation requires end to end quality of service on every bit of network to have any real meaning. Throttling can be applied in a single box anywhere along the path and will still be effective. You can use the prioritisation methods to throttle, you can't use throttling methods to do prioritisation.

Disagree. Throttling is equally ineffective without network-wide provisions to throttle traffic where it actually matters. Unless you're implying throttling traffic based on a fixed rate cap and to a level that is so low that it is far lower than any other potential bottleneck on the network. Which is not the type of deprioritisation throttling I'm talking about. Note that I said prioritization and deprioritization. Which uses the exact same mechanisms.

Also on most consumer networks there is only one main bottleneck and prioritization only needs to happen at that one location to be effective. It does not need to be network wide.

But what is stopping Voda or O2 also prioritizing Root Metrics traffic to "assist" the results?

Maybe EE are doing this, maybe they're not... but it's not as if they're the only network *capable* of doing so...

I sense sour grapes.

I have sensed this all the way through, looking to pick a hole wherever possible, "it's the phone's they're using", "it's the area they're picking", "could they be traffic priority" (despite no evidence).

Whilst I accept it's not perfect, I haven't seen anything that discredits the report or it's value as a general guide.

I have zero trust in their reports, its just far too convenient for me to believe when money gets transferred from EE to RM.

especially as my own personal experiences are the exact opposite of what they report.

The one thing that's true of any data, is that you can present it in a multitude of stories to fit the agenda you want.

Just downloaded and installed the RM app and compared its resulsts to the app i normally use

RM - 3.37Mbps DL & 2.61Mbps UL
http://i.imgur.com/WsPpemj.png

v

Traffic Monitor Plus 5.62Mbps DL & 2.81Mbps UL
http://i.imgur.com/QMJ9tYI.png
Both results taken using my Note 3

Tried a couple more and the RM app is always about 2Mbps slower, looks to me like RM manipulate the results downwards.

You can write a report or PR release cleverly, but when all networks are measured and all the metrics measured are listed and reported then you are just comparing like for like.

The fact that EE buy some of the detailed reports from Root Metrics isn't really a factor, Root Metrics are independent and are purely a research company.

They measure 'goodput' not throughput, unlike most speedtest sites. I see once again jumping to conclusions or looking for a conspiracy, regardless all networks would be measured using the same system.

"Goodput" would vary the results by an insignificant amount - 2-3% and usually impossible to determine using the RM app when it rounds to the nearest whole megabit. It certainly doesn't account for the 67% difference in his results.

It is one reason for slower speeds, it is a fair test if the same servers are used for all network tests. Goodput could have more of an effect on mobile connections as you can see a lot of packet retransmission. I'm not going to add much weight to one paranoid user sitting in their armchair running different speed tests. The user has already listed several things and is sitting there nitpicking.

I find Rootmetrics is it actually slightly slower than the Speedtest app (if you are using a decent server on Speedtest i.e. London Structured Comms, Fidonet, Namesco as the local servers cannot usually handle Fibre/4G speeds)

I see many folk using crap servers on Speedtest.net.. Only use the above or Preston if you are nearer to it.

Actually the Amsterdam Ookla server is the best as it has specially tuned connection parameters for fast connections (specifically much larger receive windows for upload tests and much larger download file sizes).

However I almost always use the London ones you mention and yet always get consistently lower speeds in the Speedtest.net app.