Data generated when playing games over PSN - impact on monthly quotas etc.

Premise

I notice on dedicated gaming forums that people have wondered how much data is generated by on-line gaming. There have been claims of upwards to 200K per second being generated which could have a big impact on people's monthly quotas, etc.

So I conducted my own test, mainly because I'm helping a mate with an unrelated project that required a rough measurement of data services anyway.

Like any network-capable game, PSN requires that you sign-in to a central server and then begin transmitting data about your location in the game, etc. Likewise you receive information about the location of other players in the game. I used three games as a test base:

• Burnout Paradise: This can generate data forever - the game doesn't actually have a formal "end" until you quit.
• OutRun Online Arcade: Provides bursts of 5 minutes of online activity before everyone reconvenes in a lobby for the next game.
• Wipeout HD: Also provides bursts of activity until meeting in a lobby again.

I had three local PS3's connected via a Netgear 10/100 switch and they all gateway through a Linux Smoothwall router to the Internet.

Smoothwall has the ability to monitor the individual bandwidth consumed by each LAN device connected to it, listed by local IP address. Since I'm using a network switch instead of a hub, the benefit here is that if the PS3's were to talk to each other directly, Smoothwall would not register any data from the PS3's because they are not passing through to the Internet. If they DO talk to the Internet, then Smoothwall would register a spike in usage for the IP address associated with the device doing the talking.

The tests were simple:

• Play a 15 minute game of Burnout Paradise doing various challenges and general stuffing around in an 8-way on-line game with other players on the Internet.
• Play a 15 minute game of Burnout Paradise doing various challenges and general stuffing around in a 3-way on-line game with other players on the Internet.
• Play a 15 minute game of Burnout Paradise doing various challenges and general stuffing around in a 3-way on-line game with only my three local PS3's as the only participants in the game.
• Play various short challenges in a multi-player game of OutRun Online Arcade with other players on the Internet.
• Play various short challenges in a 3-player game of OutRun Online Arcade with the local PS3's in a private invite-only game.
• Play various short challenges in a multi-player game of WipeOut HD with other players on the Internet.
• Play various short challenges in a 3-player game of WipeOut HD with the local PS3's in a private invite-only game.

One of the goals is to measure data transmission on the Internet. It has been widely claimed that the more players attached to a game means more data generated (seems logical) and I also wanted to see if there was any constant stream of data generated whilst in the lobby of a game where there was technically no action. I also wanted to see if there was a difference (or consistency) in the volume of data generated between different games, eg: if Burnout Paradise generated data at 50K/s, would OutRun do 60K/s and WipeOut HD 40K/s?

Practical

I found that when signing onto PSN, the PS3 would spike usage to the Internet to around 180kbps for a brief second and very quickly drop back to zero. Sending a message via the XMB from one PS3 to another, data would spike at no more than about 2kbps for a split second at the time of sending.

I fired up Burnout Paradise and joined an 8-way game already in play with only one of the local PS3's I had. Contrary to other claims I've read, once connection was made and the data stream settled, the PS3 was recorded receiving no more than 51kbps and sending no more than 52kbps. When another player dropped out, I quickly joined a second local PS3 into that game and observed that it too was sending and receiving data at 51 and 52kbps respectively. This so far supports the theory that data usage doubles each time you add a machine.

I disconnected both PS3's and and reconnected only one of them to another game that only had two other Internet players - a total of three players including myself. This time, the data generated settled at 15kbps down and 13kbps up.

I disconnected and then created my own public game using one local PS3 only and waited for 7 other Internet players to connect to me. With 7 other machines talking to me, you'd therefore expect the incoming and outgoing data to be something upwards to 408kbps out and 416kbps in, right? Nope. It was no more than 51kbps in and 52kbps out again, and I was the host!

Playing various challenges and breaks in gameplay where there was idle time at the start of a race etc did not increase the flow of data, however there were some lulls in data transmission at idle times.

So I (politely) ended the open game and created a new one, this time as invite-only. Burnout Paradise requires connection to PSN at all times for any form of multi-player gaming, so at this point I could have invited someone on the Internet and they could join, but in this case, I only wanted to invite the other two local PS3's who were also connected to PSN into my hosted game.

This is where it got interesting. Sure enough, there were spikes of Internet usage recorded as the machines received the invite and accepted and then they joined. As soon as my host machine acknowledged entry of the two invitees, all of a sudden Internet usage flatlined. Zip, zilch, nada.

Initially I thought my Internet connection had dropped out or something, but no - sure enough, each machine could see the other two players on the map, and when driving up to meet each other face to face, animation was smooth, positioning was smooth (not jerky or jumping around like a bad connection does) and yet Smoothwall recorded absolutely nothing going out to the Internet. The network switch OTOH, was happily flickering away showing that a regular flow of data was indeed being transmitted between all three machines, but the link to the Smoothwall box barely flickered, only blinking every so often. Did this mean that I was effectively having a LAN-only game?

I started a few challenges, beat some road rules, etc. Each time a world-recordable factoid about a player was achieved, Smoothwall registered a tiny spike in Internet usage for a second or more, with one observed sustained transmit/receive of no more than 1kbps for a period of about 5-10 seconds, no doubt the game was transmitting the achievement to the Burnout Paradise central servers to update that player's profile. All up, the Internet was used very very little. The majority of all Internet data transmissions made came from the host PS3. Very little came from the two invitee PS3's.

If I disconnected the Smoothwall server from the network, thus severing the Internet connection for all three PS3's, within seconds the game terminated saying it lost connection to the game servers.

When allowing an Internet 3rd party to then join the private game between the three local PS3's, all three PS3's suddenly started communicating the same amount of data to and from the Internet, ie: in a 4-way game, all three local PS3's went from transmitting zero data to 35kbps/20kbps each, a total of 105kbps/60kbps.

All up I recorded the following data rates for various player numbers in Burnout Paradise:

Code:

Players   Data IN   Data OUT
1         N/A       N/A
2         7kbps     7kbps
3         15kbps    13kbps
4         35kbps    20kbps
5         45kbps    20kbps
6         47kbps    36kbps
7         48kbps    40kbps
8         51kbps    52kbps

Audio conversation added roughly 2kbps on top of each figure shown. Some of the above figures will be slightly skewed due to it being difficult to to find a game where no-one was talking to see the usage.

So I ended Burnout Paradise and fired up OutRun Online Arcade. Without going into extreme detail, the results were pretty much identical to Burnout Paradise. On-line sustained play activity netted usage no greater than 60kbps both ways with a full complement of players and idle time in lobbies was met with near-zero usage with small spikes every time a player changed a parameter about their vehicle. Occasional spikes in usage went no higher than 108kbps, typically at the start of a new game. Local PS3 play exhibited the same as before as well - near zero activity on the Internet with the majority of all activity being kept between the three local participant PS3's only.

Wipeout HD - same thing. Same data levels during play with other world players, idle lobbies, and next to no data being transferred when playing with the local PS3's.

Conclusion

• Data usage increases the more players who are connected - confirmed.
• Data usage does vary between games, but not by too much. ie: usage for 4 players in Burnout Paradise was not identical to 4-players in OutRun or Wipeout HD, but they were similar within 20 or so kilobits.
• Overall settled usage never really exceeded 52kbps on average. Since most games appear to max out at 8 simultaneous players, allowing for some overhead between games, it's probably safe to assume that you will never exceed 60kbps in most games in general play. In my tests, this means that for continuous Internet play I eat up no more than about 23MiB per hour of my quota (60kbps = 7.5KiB per second, or 450KiB per minute, or 26MiB in an hour). If your uploads also contribute to your quota, then just double these figures.

For everyone who has always complained about there being no specific off-line LAN game modes in games, you now have some evidence that supports there being next to no Internet activity for all your locally-connected PS3's (as long as you don't include any Internet players). PSN appears to be smart enough to have each machine communicate directly with each other rather than constantly pass through a central server for everything, so whilst playing a multi-player game for the most part still requires you to HAVE an Internet connection to login to PSN initially, at least you know that you won't be chewing up truckloads of bandwidth in the process of gaming.

This test has obviously not been completely scientific, however I think the results are more than concrete enough to have illustrated the point. I'd be interested to hear from anyone else who has conducted their own network tests.

 

whoa awesome stuff!
next time the mrs complains about me laggin her facebook im going to point her right here

 

Impressive work! Now the only thing you have to worry about is some arse stealing your work and posting it under their name somewhere else.

Can I suggest you edit the post and add a "please do not reproduce without permission" somewhere in there to at least attempt to slow down rampant internet douche-baggery?

 

Excellent stuff.

 

I'm not entirely fussed. If I wanted to make a big deal about attribution, I'd have posted this up on my blog, not here.

Besides, any disclaimer text is easily removed/replaced and you could easily compare posted dates on the plagerised copy anyway.

EDIT: I just realised that this was my 3,000th post! Woohoo!

 

True, it's just that you put a lot of effort into it and it's a very nice piece of work

 

granted this is oldish data BUT

back in the days of halo 2, a guy i used to play with ran a box on his works 100mb connection, we would join its party and take the idle player into games - sure you were a man down but were assured of host over people OS.

anyway - he did some looking at it and even with the connection getting flogged basically 24/7 usage was basically under 100mb/day if i remember correctly....