New potential shared host arrangment

[macsforme]

Hi all,

Many people (myself included) have desired to host professional-class BZFlag servers, but have not had the means to obtain a host with enough bandwidth, etc. Some of these people run their servers on home connections (which in many cases is against the terms of service from their ISP), but run into trouble with connection reliability, and often have high lag and/or high jitter problems. Also, most home connections have very limited bandwidth to begin with, and have enough capacity for 10-15 players.

The obvious alternative is to pay $70 - $150 per month for a dedicated server. However, since many of these people are students or young kids, they do not have the financial resources to support such a server. Even those who can pay the amount are often not interested in such a high monthly payment for a computer game server. It is, after all, just a game.

Some people have managed to find others willing to host their servers, while other people have been able to find a company or school who is willing to donate the resources. While in an ideal world we would all be so lucky, when it really comes down to it everything in this world comes with a price.

Because I believe that having high-quality servers for BZFlag is important, and I would like to make hosting possible for as many people as possible, I am considering heading up a shared host with good speed and capacity on an extremely fast and reliable connection. I was considering purchasing dedicated server hosting for business and personal reasons (very light use), and this would allow me to allocate all the extra bandwidth to people who would like to use it for their BZFlag servers.

Please note that this is not a commercial endeavor of any kind; all users would be sharing in the cost of renting the server on a fair and equitable basis. The server would be configured so that resources are allocated fairly to all users. Users would also have to agree to reasonable terms of service so that the rights of the other users are not compromised.

The particular server I am looking at right now will have an average monthly capacity of 2.5mb - 3mb continuous bandwidth available (although the actual connection capacity at any given moment will be several times as much). For general-purpose users, I've worked out a preliminary monthly pricing plan, which will include:

2GB storage space
4 network ports for bzfs (more available at small additional cost)
Apache virtual host configuration
Player slot allowance at rate of $0.50 per slot up to 20 slots (special arrangements available for more slots)
100GB monthly transfer limit

We may chose to do virtual private servers (VPS), in which case settings would be slightly different. The per-slot monthly cost may vary slightly from this if/when this gets underway. As you can see this plan is mainly targeted at the smaller (currently home-run) servers (although more popular servers are also welcome). For users with special needs and/or very popular servers, custom arrangements will be available.

I will be designing/building/configuring the server, settings, info website, control panel website, etc. Measures will be put in place to ensure that users do not pay for downtime that was caused by software or configuration problems. Users will have to take certain precautions to safeguard their accounts against malicious usage. Payment will be set up by PayPal, and I plan to automate a system so that account openings/closures/lockouts (due to non-renewal) will be as automated as possible.

I'm quite interested in hearing people's reaction to this idea. I think this will be a very useful setup for server owners who can't afford or have no desire to obtain a dedicated host. For most servers, this will be a very small monthly payment (~$10) for a fast, reliable host. I am working with the BZFX and Plosileague team regarding the possibility of providing hosting for these servers, so interest is already coming in.

Thanks!

[tadd]

I have pondered the idea of making a bzflag server self sufficient.

I think that some bzflag hosters would be willing to commit about $50/year for a shell account. I don't really think that it would be worth more than that. The only real limitations imposed on the users would be that they could not use bandwidth higher than the fraction that the server and network connection can permit, and they can't run programs beyond what is supplied by the server owner, and thus can't drain CPU cycles more than bzfs.

One thing (for good or bad) about bzflag is that there are no more than a thousand or so players in the entire world at a given time. It is unlikely that all of the players combined would overload a well connected server. I suspect that no real bandwidth limitation needs to be imposed, so long as the only service offered is bzfs. Web service exposes the server to monsterous load if the users are given no limits on what they can serve. There are enough hosters that give away free web service that I don't see offering web is necessary or worth any real value to the users.

Charges for the server will stay the same month after month so long as only bzfs was running. Over-usage of the network connection is not a problem.

The entire cost of operation of the server is limited to amortization of the cost of the computer, cost of the colocation, and a cushion to cover emergencies = about the cost of the whole server so that you are covered for shipping and tech support for server replacement etcetera.

So.. if the server costs $700 including the OS, and lasts 3 years, with networking costs of $80/month, your total cost of operation for 3 years is $1400 (equip) + $2880 (connection) = $4280 divided by 3 years = $1425. If you had 28 shell accounts at $50/year you break even (even with the extra cushion money).

I am in doubt about the cost of a colocation with 5megabytes/second continuous bandwidth at the $80/month pricepoint. I can't find any examples at this moment and don't have any more time to look. I'll edit this once I get real figures. However, 5megabytes/second is the maximum bandwisth i can imagine with all 1000 worldwide players on at the same time. I think we can probably count on the average usage working out to far lower than the basic hosting offered for the named figure.

The trick is finding 28 people who will pony up the bux in advance. If I can't find that, I can't have a server.

This is substantially less than your initial proposal, Constitution, and even at $50/year I don't know if this plan could attract 28 server ops.


At this time, I run my server for free but with huge limitations imposed on the customers.


So... survey question. Would you pay $50 for a shell account with no limits on # of players? I'd put a limit of maybe 10 or so maps per shell. heehee.
Tadd

[CannonBallGuy]

Tadd, about your pricing....
The MoFo group pay around $70/month for a dedicated server with more than enough resources for all our servers *and we probably host a larger proportion of the servers listed than you realise).
There are around 7 active users who "donate". Personally, I do my upmost to pay $10 each month to the dude that actually pays the monthly bill.

However, the "hosting plan" proposed by Constitution ($10 for 20 slots) would bring in as much money per user as our setup, but with a potentially lower limit on resources.

What I'm trying to say is, at 20 slots per user, he could get more than 10 users going at $10/month. In theory, he could make a profit on this.

[A Meteorite]

One thing (for good or bad) about bzflag is that there are no more than a thousand or so players in the entire world at a given time. It is unlikely that all of the players combined would overload a well connected server. I suspect that no real bandwidth limitation needs to be imposed, so long as the only service offered is bzfs. Web service exposes the server to monsterous load if the users are given no limits on what they can serve. There are enough hosters that give away free web service that I don't see offering web is necessary or worth any real value to the users.

I am in doubt about the cost of a colocation with 5megabytes/second continuous bandwidth at the $80/month pricepoint. I can't find any examples at this moment and don't have any more time to look. I'll edit this once I get real figures. However, 5megabytes/second is the maximum bandwisth i can imagine with all 1000 worldwide players on at the same time. I think we can probably count on the average usage working out to far lower than the basic hosting offered for the named figure.

5Megabytes/second can not sustain 1,000 players. It just won't happen (bandwidth on one server scales exponentially).

As to "web services exposing the server to monstrous load" - that is why you design the service correctly. You can set-up accounts that can only turn on so many servers. Easy as pie.

So... survey question. Would you pay $50 for a shell account with no limits on # of players? I'd put a limit of maybe 10 or so maps per shell. heehee.
Tadd

Erm, this is Constitution's thread about hosting on his machine. Not your idea of doing shells for $50/year. On that note, please don't hijack threads with your stuff. It belongs in your own thread. It ain't cool (and doesn't make your ego look very good either).

[macsforme]

What I'm trying to say is, at 20 slots per user, he could get more than 10 users going at $10/month. In theory, he could make a profit on this.

Although I would like to serve as many users as I can, I'm not anticipating getting a ton of users (at least at first). I realize that you were speaking in theory, but I am really not interested in making a profit on this. If I had enough users that there was extra money, I would put it back into the venture, either investing in more bandwidth, etc., or discounting the rates for my hosters.

At this time, I run my server for free but with huge limitations imposed on the customers.

I'm glad that some people are able to host server for free, but I personally don't have the resources to offer such a service. This initiative is about bringing professional-level hosting services to people at as low a cost to them as possible. If there are free alternatives that are available across-the-board that can bring the same level of service (which I do not believe there are), then we should take that route instead of this one.

Also, running a server with no limitations on the number of server slots is unlikely to work well. Just imagine getting two or three servers as popular as badgerking's, BZFX, In-League, etc. on the box. Bad news.

This is substantially less than your initial proposal, Constitution, and even at $50/year I don't know if this plan could attract 28 server ops.

Not necessarily. As I've said, I haven't ironed out the final details yet (there will probably also be several different kinds of plans, for instance per-slot, per-bandwidth, per-server, etc.) Even given my original details, a $50/year plan would be feasible. It's important not to overcommit server resources so you don't end up short-changing people.

The trick is finding 28 people who will pony up the bux in advance. If I can't find that, I can't have a server.

I'm not asking people to put up such a large amount of money in advance. Payments would be on a month-to-month basis (among other methods of insuring that people's investments are protected). This is the internet. Sure, people can build up good reputations just like in real life; but in my opinion, people should not be asked to put up a lot of money for an internet deal with someone they hardly know.

[that exploding tank]

I'm not asking people to put up such a large amount of money in advance. Payments would be on a month-to-month basis (among other methods of insuring that people's investments are protected). This is the internet. Sure, people can build up good reputations just like in real life; but in my opinion, people should not be asked to put up a lot of money for an internet deal with someone they hardly know.

This sounds like our public radio station in their spring membership campaign. Problem: the radio system ran short on pledges. Bigger problem: the minimum that you had to commit in order to pledge is $35/year. ($25/year for students and seniors). And believe me, these people were spending a full 15 minutes each hour on a statewide band to get pledges in. I'm feeling that people will not want to pledge their $50/year in order to support a gaming server. Besides, it kind of goes against the principle of BZFlag, where everything offered here is absolutely free, with the possible exception of server maintenance costs. Furthermore, there really is no way to effectively advertise this, and if I am thinking right, it is against the forum rules to advertise for any service here that costs money.

It's a good idea, but I don't think that it's practical. We're not that cohesive a group anyway; we just play the game.

[tadd]

Erm, this is Constitution's thread about hosting on his machine.

please don't hijack threads with your stuff.

I was not talking about 'my stuff'. I was trying to point out that I support the free model but understand that Consitution does have a leg to stand on since (as a free server op) I can and do impose such huge limitations on my 'customers'. I don't think it is hijacking to try to help narrow constitution's idea down to something I think would be workable.

That said, after considering the good comments that have come into this thread since:

Besides, it kind of goes against the principle of BZFlag, where everything offered here is absolutely free, with the possible exception of server maintenance costs.

I agree with that exploding tank. Well said.
Free is good. There are minimal obligations that way.

Just imagine getting two or three servers as popular as badgerking's, BZFX, In-League, etc. on the box. Bad news.

I'm really hoping to see a real demonstration of this problem. 2tanks and missile war are doing a pretty good job of being crowded these last couple of days. I have been watching lagstats for some players whose latency I know well, and I haven't seen any impact at all with having 60 players on my Mac Mini at once. Perhaps if there were 120?

Tadd

[CannonBallGuy]

This sounds like our public radio station in their spring membership campaign. Problem: the radio system ran short on pledges.

If someone doesn't pay one month, their server would be terminated until payments resume.

Bigger problem: the minimum that you had to commit in order to pledge is $35/year. ($25/year for students and seniors). And believe me, these people were spending a full 15 minutes each hour on a statewide band to get pledges in. I'm feeling that people will not want to pledge their $50/year in order to support a gaming server.

Nobody would need to pay $50 at once...

I'm not asking people to put up such a large amount of money in advance. Payments would be on a month-to-month basis (among other methods of insuring that people's investments are protected). This is the internet. Sure, people can build up good reputations just like in real life; but in my opinion, people should not be asked to put up a lot of money for an internet deal with someone they hardly know.

Besides, it kind of goes against the principle of BZFlag, where everything offered here is absolutely free, with the possible exception of server maintenance costs.

Dude, are you daft? You think those 200+ servers on the "Find Server List" are free? You don't think anyone has to pay for those out of their own pocket?

Furthermore, there really is no way to effectively advertise this, and if I am thinking right, it is against the forum rules to advertise for any service here that costs money.

What rule, exactly, would be violated?

[tadd]

This sounds like our public radio station in their spring membership campaign. Problem: the radio system ran short on pledges.

If someone doesn't pay one month, their server would be terminated until payments resume.

I think the point was that the server op would run out of money if there weren't enough customers.

Dude, are you daft? You think those 200+ servers on the "Find Server List" are free? You don't think anyone has to pay for those out of their own pocket?

I suspect that some of them are on university machines, and some served via home internet connections.

[sbgodin]

I suspect that some of them are on university machines, and some served via home internet connections.

The example for one of them:
- I'm the admin
- Tuxfamily (Free Software assoc.) has the cpu and ssh access
- Frontier is a hoster that allows Tuxfamily to use its bandwidth

[CannonBallGuy]

I suspect that some of them are on university machines, and some served via home internet connections.

The example for one of them:
- I'm the admin
- Tuxfamily (Free Software assoc.) has the cpu and ssh access
- Frontier is a hoster that allows Tuxfamily to use its bandwidth

In those cases, bills still have to be paid... It's just not you that is paying them.
In which case, great. If you have access to free hosting - use it.
If not, maybe Constituion's idea is for you...

[L4m3r]

I'm really hoping to see a real demonstration of this problem. 2tanks and missile war are doing a pretty good job of being crowded these last couple of days. I have been watching lagstats for some players whose latency I know well, and I haven't seen any impact at all with having 60 players on my Mac Mini at once. Perhaps if there were 120?

Lag and bandwidth consumption aren't exactly linear in nature when the number of players increases. Because the server has to send data for all players to each player, it actually goes up quadratically on each server- i.e., with the same settings 20 players will use about four times as much bandwidth as 10 players, on one server. This is why large games can be difficult to host. However, when you've got a lot of players split up over multiple servers, the consumption is a lot less because you're not sending as much data around.

BZFS hardly uses any CPU, so that's rarely an issue. Your mini is on a fast pipe, so I think you'd be hard-pressed to flood it enough to impact performance. Most likely, you'd be flooding players' connections (too much info to one client for most to handle) before you had a server-side problem.

Of course, bandwidth usage is also affected by the number of shots, how often people jump, and a whole lot of smaller factors. In the end, though, total bandwidth usage is almost always the limiting factor on a "professional grade" BZFlag server, not the immediate speed of the pipe.

Anyway, that said, I don't see how this conflicts with the free nature of BZFlag. Good servers aren't free for most people. I see this as a way to share the expense. Sharing a server allows each participant to rent exactly what he/she needs, a fraction of the entire box, for a fraction of the price. A dedicated server is usually overkill for one person, and there aren't a lot of good "smaller" options out there.

[macsforme]

If someone doesn't pay one month, their server would be terminated until payments resume.

I think the point was that the server op would run out of money if there weren't enough customers.

I'm not running into something like this without a good deal of forethought. This initiative, like any other like it, needs to have a solid base of resources before launching. If I do get this going, I would do so only if and when I was properly prepared for the worst-case scenario.

Besides, it kind of goes against the principle of BZFlag, where everything offered here is absolutely free, with the possible exception of server maintenance costs.

Dude, are you daft? You think those 200+ servers on the "Find Server List" are free? You don't think anyone has to pay for those out of their own pocket?

Thank you CannonBallGuy... I agree that every server is ultimately paid for by somebody, and that perhaps this is not a vivid reality to a large portion of our player community.

that exploding tank, thank you for your feedback. The very reason why I posted this idea to the BZBB was to get some feedback on how many people would be interested in something like this. Until that feedback actually comes in, I'm keeping an open mind.

[trainee]

The BZ open source/freeware criterion, TTBOMK, applies to LGPL licensing (code, distribution, copyright) and a prohibition for charging players to play.

While the BZ client is freeware, and the servers, for the most part, are unrestricted, some players might be willing to contribute to the operation of specific servers. This might be particularly applicable to members of teams or players that are committed to the operation of a specific server and that have the desire and financial ability to contribute.

Sourceforge, which is not a tax-exempt entity, accepts donations to assist in providing support for their sponsored projects, and there does not appear to be any restriction regarding BZ players' contributions, whether or not such support is directed to a particular server.

that exploding tank's stated that " ...the principle of BZFlag [is] where everything offered here is absolutely free... ." IMO, players' voluntary contributions are consistent with that exploding tank's statement because voluntary contributions could help provide and maintain the hardware and bandwidth that enable the game to be enjoyed freely.

I don't have the ability to develop a map or the knowledge or financial ability to operate and maintain a server, but I am willing to contribute financially to the operation of a server or server.

After all, it's "good clean fun... sorta."

[L4m3r]

The BZ open source/freeware criterion, TTBOMK, applies to LGPL licensing (code, distribution, copyright) and a prohibition for charging players to play.

LGPL does not prohibit charging to play or even charging to distribute. It only requires that the source code be easily available- i.e., BZFlag could be sold retail for a price. However, the LGPL requires that the source code be provided, and there's nothing to stop someone else from posting a build and distributing it for free (as long as the source is still provided). Such is the case with XChat, for example.

And anyway, the fact that a rented server host runs BZFS does not affect its legitimacy. You're not being charged for the software, but rather for the bandwidth and maintenance. It could be compared to, say, the fact that most of us have to pay for an internet connection so that we can play.

I don't think the intention here is to charge players, but rather, to charge people who run servers on this shared host. No one is being forced to do do anything, but it's not necessarily "voluntary contributions" so much as a rented service.

There are no ethical or licensing issues here. It's all about the money... namely, whether or not enough people are interested for the server to be a practical operation.

[JeffM]

From what I can see by the first set of posts here, None of them are even suggesting pay to play. There are talking about pay to host, and every server out there is pay to host by someone (internet ain't free).

It looks like once again somebody didn't read all the words and overreacted, and the thread got off topic.

1) they are talking about paying to host. Aka splinting up the cost of the server over a number of server owners/admins. This is happening right now on other servers. It has nothing to do with the software license. The player "slot" thing is just a way to split up the bandwidth used by the server. The people that pay would be people that want to run a server( pick a map, rules, admin it, etc... ) on there own port/name/map

2) as L4m3r said, the LGPL has nothing to do with money or being "free". BZflag is not "Freeware" it is LGPL and distributed for free by the project via sourceforge. The LGPL license only has to do with the source and the distribution of the software. End of story. Since that license ONLY deals with the distribution of the software, NOT the content or servers it connects to, a server owner can do pay to play if they want. It is totally within the license of the software. Just as you can use the open source firefox web browser to go and subscribe to a website that charges for content, and the application still remains open source.

Personally I think a nice pay to play server could do quite well if it was made reasonable (2-3$ a month ). The money can go to pay for a nice fast server, pay for active administrators, and in some way would help prevent cheating ( who would pay someone to cheat and get banned? ). It would just need some management.

There will always be free servers, but that doesn't mean they will be the only servers. Everyone needs to understand the separation between software and service, and just get over it. If you don't like the concept, fine just don't pay to play there or pay to host, do your own thing, that's the beauty of open source

[trainee]

Great. I'm willing to share some costs. (I don't know diddly about computers, however, so I can't contribute much there.)

[QuantumBeep]

Pay-to-play would have to be very, very good to be worth money, and anyone attempting to start such a service would probably get treated really badly by everyone who didn't feel like paying.

2mb-3mb isn't enough to host more than one very active game. Mofo was able to saturate a 10 mbit connection (granted we were at the time hosting 3/4 of all the players everywhere - bet you didn't know that!). Now we're on a 100mbit LAN pipe.

[macsforme]

Mofo was able to saturate a 10 mbit connection (granted we were at the time hosting 3/4 of all the players everywhere - bet you didn't know that!).

Woah! That's a lot of bandwidth!

Anyway, thanks all for the reactions... I'll continue contemplating the idea, and see if it can be made to work. More comments/reactions would be appreciated, as always.

[tadd]

2mb-3mb isn't enough to host more than one very active game. Mofo was able to saturate a 10 mbit connection (granted we were at the time hosting 3/4 of all the players everywhere - bet you didn't know that!). Now we're on a 100mbit LAN pipe.

that's great info QuantumBeep. thanks. I was wondering what a real-world bandwidth utilization would look like.