VDSes, Full machines, and Virtualization

[Agent of Corruption]

Hello,

Our community/organization/whatever currently utilizes one 6-core managed linux VDS host for 6 Garry's Mod servers of varying player count and resource usage. We have another 2 single-core linux VDSes instances for web-hosting and related concerns. We've currently hit in an obstacle in attempting to run an Arma 3 server or two as NFOservers does not allow putting an Arma 3 server on a managed linux VDS (See below). The leads to a few alternative options. We can switch to windows, pay an extra $15 a month, and deal with a bit more overhead and some other annoyances of windows, we can spin up another VPS either Windows managed or expand one of our two linux, or we can take advantage of the sale on free setup for full machines and rent one and ideally use some kind of hypervisor whether it be Xen or VirtualBox or anything else NFO allows to try and condense all our VDSes and resources into a couple VMs running the Garry's Mod servers, an Arma server or two, and all our varying web projects and other python stuff, and node.JS, and etc. However, I don't know how effective a quad core CPU would be at efficiently running all those things inside virtual machines as my experience comes from using things such as ESXi, proxmox, and Hyper-V to virtualize windows Active Directory and linux web-hosting infrastructure, not game servers.

The reason I'm asking you guys is because we're just trying to make efficient use of limited resources and none of these solutions seem ideal. We want to take full use of the resources of a 6-core VDS and would rather not pay $15 more a month ($180 a year) for the pleasure (and extra overhead and problems) and if we create or expand another VDS, it's more money for more resources that we don't need. If we go to the full machine route, we don't have all the advantages managed servers provide in terms of both convenience and performance.

In an ideal world, the most preferable option would be if NFOservers offered more games for managed linux. With it being $15/mo more to have a windows VDS for no real benefit compared to Linux, it would be nice if NFOservers supported more games like Arma 3 and Rust in their managed platform which perform fine on linux. Many of the games on this list https://gameservermanagers.com/servers/ (LGSM is a software for easily updating/running game servers on linux) are games that NFO supports... but only for windows. Unfortunately, it seems like the continued response is that it isn't a priority because there is no perceived demand. That seems a little odd considering I think there would likely be a lot of demand to save $15/mo for windows licensing which in many cases isn't needed.

We'd appreciate any advice on the matter of what route would be the most effective in terms of cost and performance, or whether to wait and hope managed linux support is increased.

Thanks!

[TacTicToe]

Unfortunately Linux is a vast minority in usage. Probably less then 5% of the gaming population is on Linux. It is why so few game developers port to Linux, it is just not cost effective. Same with Mac. We use Linux for our web servers as it works very well. We hosts a few game servers on Ubuntu using LGSM. Bottom line, it isn't personal, its just business. The demand for Linux gaming *anything* is just so low it isn't financially viable for developers or GSP's to give it serious consideration.

Windows works extremely well for hosting game servers, especially when the game developer actually spends the time to optimize their software. Windows is the dominant OS platform for hosting servers and most likely will be for the forseeable future.

https://www.vg247.com/2013/03/04/steam- ... t-popular/

https://www.quora.com/Video-Game-Indust ... -Mac-users

Old articles, but makes my point. Even when I look at he stats for visitors to our website, the Linux users is always well under 5% of those connecting to our site.

I would suggest going with unmanaged hosting or better yet a full machine. A full Windows machine would allow you to host ALL the servers you want and no overhead. It is what we use and has always worked great.

As we host using both Linux and Windows, I can tell you unequivocally that the benefits and ease of use with Windows blows the doors off Linux, hands down.

[Edge100x]

With many games, the Linux version does play second fiddle to the Windows one. Lower performance, higher resource usage, more bugs, and so on.

Windows doesn't have notable overhead that you have to worry about. Just a bit of extra memory -- maybe a few hundred MB. The main downsides to Windows are its licensing fee and the fact that Linux is more compatible with many webserver-related applications.

If you do buy a dedicated machine, I'd recommend against virtualizing it, for performance reasons. If you want to use Windows for the game servers and Linux for the webserver, a small standalone VDS for the Linux parts might be a good idea.

We haven't heard much demand for more managed Linux game server options, but we do plan to add further support in the future. Thank you for your feedback specifically on Arma 3.

[TacTicToe]

I think the Windows overhead he is talking about is the negligible 3-5% you may get on a VDS due to Xen. His other gripe was the $15 upcharge for using Windows.

I still think they would do better just to get a full box unmanaged with a Linux distro if that is what they really want. Having full use of all the resources like that is just awesome.

Personally, I would prefer to see an API for collecting donations from people as a priority. It is something we have been asking for, for years, and is still sitting on the back burner. John only has so much time to spend doing all the things he does.

[Edge100x]

One sad fact about donations is that they cost us extra money to handle. We don't pass that along to customers, but we pay a significantly higher amount in fees for a $50 server paid as ten $5 donations than as a single $50 payment, for instance. If we become more like a real payment processor and spend the time and money to add more donation-oriented features, we will also encourage their use and increase the fees we have to pay. This, in turn, could mean being forced to do what payment processors do and implement a fee -- or, raise the prices of our services further to subsidize the increase. So, enhancing donation support is one of those actions that is not a slam-dunk win-win.

Should we still improve it? Yes, I believe so, as part of a more general API. But, its initial and continuing expense factor into its priority.