Minimum Recommended Server Specs for rFactor 2

[MaXyM]

obviously this is the server sending info on the cars to the other cars. The more cars, the more sending out. It's exponential growth, not steady growth.

Yes. thats why I multiplied base data traffic needed for single car per power 2 of number of cars (9,6 * 30 * 30).
So it is clear I'm talking about exp growth. But still there is data overhead missing. Or maybe there is something wrong with prediction of data traffic for 10 cars.

[Tim Wheatley]

Yes. thats why I multiplied base data traffic needed for single car per power 2 of number of cars (9,6 * 30 * 30).
So it is clear I'm talking about exp growth. But still there is data overhead missing. Or maybe there is something wrong with prediction of data traffic for 10 cars.

It isn't a prediction, it's a detected test and monitoring. That's why these should be treated as minimums.

[Graag]

Is 30 drivers maximum or can server handle more? We used 40 driver maximum with rFactor.

[Tim Wheatley]

Is 30 drivers maximum or can server handle more? We used 40 driver maximum with rFactor.

No, but we haven't yet been able to ascertain the numbers above that amount. We might be able to finalize it and then give you those answers during the open beta... This is also covered partly in the upcoming Q&A which should be posted by the site which has it on Nov 1st.

[LesiU]

Nice

Tim, does ISI have internet connections that could hold some big events, like for 60-70 people (or even more, if track allows)? Maybe 2-3 such events during open beta period? That could not only tell, what rFactor 2 is capable of but also ISI stuff would have more data about real world server requirements and work on optimal dedic settings?

[Tim Wheatley]

Nice

Tim, does ISI have internet connections that could hold some big events, like for 60-70 people (or even more, if track allows)? Maybe 2-3 such events during open beta period? That could not only tell, what rFactor 2 is capable of but also ISI stuff would have more data about real world server requirements and work on optimal dedic settings?

We'll probably host some stuff during the beta. We do now with the closed one.

[LesiU]

Great, good news

[Derek Speare]

Thanks for this information - will there be commercial hosting services with whom I may arrange to have a rF2 server hosted in my behalf?

[Kiro]

4MBps (0.5Mbps) upload here.
I can double that for more $$ but still wouldn't be enough by the minimums listed.
I WAS going to say "I guess I'm done hosting" but, we will see when it comes out.

[Lazza]

Hmm. I'm not sure about initial calculations.
Tim, you said that for 10 cars it is about 96kbps.
It means that data amount to send containing info about single car is about 9,6kbps.

It means that data stream of 30 cars should be 30*9,6 kbps = 288 kbps. Sending for 30 cars results in 8640 kbps / 8,6Mbps.
What data is contained in additional 14Mbps overhead?

Yes. thats why I multiplied base data traffic needed for single car per power 2 of number of cars (9,6 * 30 * 30).
So it is clear I'm talking about exp growth. But still there is data overhead missing. Or maybe there is something wrong with prediction of data traffic for 10 cars.

No, something's wrong with your calcs there Maxym.

If you want to treat it exponentially, and let's bear in mind the 'requirements' here are pretty sketchy, and there's also no reason to think it's completely exponentional in this fashion... but...

If 10 cars require 96kbps, you could bring it down exponentially via 96 / 10 / 10 = 0.96 (per player per player).

For 30 players, 0.96 * 30 * 30 = 864kbps, for each user.

864 * 30 = 25920kbps, or 3240kBps, or 3.16MBps. (MiBps if you want to differentiate 1000/1024)

[MaXyM]

If 10 cars require 96kbps, you could bring it down exponentially via 96 / 10 / 10 = 0.96 (per player per player).

No. You are wrong at this point.
The data traffic is given per client. Which means 96kpbs / 10 = 9,6kbps is amount of data generated by single client. In other word, server will generate 960kbps stream for 10 clients connected. Your calculation would be correct just for 960kps (not 96)

It is approximated value because you don't have to send data back to the client which does send data to the server. Stream also contains data not related to car movement etc.

Interesting is, that 9,6kbps is aprox the same number as required for rf1. I'm sure additional traffic is required for transfer track surface changes (caused by any single car). but still it is too large difference.

[LesiU]

EDIT: Nvm

[Lazza]

My results are quite close to the quoted figures, which are based on actual tests. Yours are not. What does that say?

[MaXyM]

So give us maths you have used.
Currently you quoted me and wrote comment with [your] mistake. Just check 1st post in this thread to get know what traffic Tim was referred to.

What does that say?

For example it might say, that ISI measured traffic for 10 cars and has calculated other results, making the same mistake you did

[Tim Wheatley]

Thanks for this information - will there be commercial hosting services with whom I may arrange to have a rF2 server hosted in my behalf?

I've already seen some places saying they are going to. That was before these specs though

[Tim Wheatley]

So give us maths you have used.
Currently you quoted me and wrote comment with [your] mistake. Just check 1st post in this thread to get know what traffic Tim was referred to.

For example it might say, that ISI measured traffic for 10 cars and has calculated other results, making the same mistake you did

There wasn't calculations, as I said, just measurements and a cutoff for best/worst. For example where it says 10-20 you might think you can host 20, but it might not be possible depending on certain conditions (time of day, track info, tire info, etc). So you might just manage 10... The bandwidth and car numbers are based on testing with real data... You can't 'work it out' unless you first account for every difference minimum and maximum in terms of data and right now if you're not testing you don't have that.

[Sp33dlink]

Woow that are huge specs..
Is there maybe a way to bundle the car location data in a way it only needs the 768kbps once?

I thought about broadcast packets in a vpn connection.
Or a switch which just copies the data for every client.

[Derek Speare]

I've already seen some places saying they are going to. That was before these specs though

Thank you, Tim - your active discussion is most appreciated! I'll be curious to know who finally shakes out when it's under way.

Derek

[MaXyM]

There wasn't calculations, as I said, just measurements and a cutoff for best/worst.

It's OK to me. I'm just curious why there is such low traffic for 10 cars since too high traffic for 30 cars.
Just asking for info about the cause. Because it is clearly related only to number of clients connected.

[niscalo]

Linux-server supported? Wow, huge specs there!