Jump to content

Forums Announcement

Read-Only Mode for Announcements & Changelogs

Dear Survivors, we'd like to inform you that this forum will transition to read-only mode. From now on, it will serve exclusively as a platform for official announcements and changelogs.

For all community discussions, debates, and engagement, we encourage you to join us on our social media platforms: Discord, Twitter/X, Facebook.

Thank you for being a valued part of our community. We look forward to connecting with you on our other channels!

Stay safe out there,
Your DayZ Team

RikkAndrsn

Need feedback from dedicated box hosts

Recommended Posts

Hey all,

Just checking in on the DayZ server stuff. I'm the president of my uni's gaming club and we've recieved several requests from our members to host DayZ and I've already gone through the requirements and we meet them all in terms of hardware. Our server is operated as a kind of gift from the comp sci department, them giving us free rack space and internet access since just about every member of the game development program is a member of our club. Our internet connection is fast enough and we exceed the specs needed but I still have a few questions.

1. We currently run a 32 person TeamSpeak 3 server, 24 person TF2 and CS:S servers, and a 20 person Minecraft server. We generally go through about 150 to 250 GB of transit a month. How realistic is DayZ's 2 TB a month figure? A jump from 5 to 10 GB of traffic a day to over 60 would turn some heads as the network is, after all, intended primarily for research.

2. ArmA II is known for server side stability issues. DayZ inherently isn't going to be free from those issues. How often do servers go down?

3. Does DayZ play nice with other apps? It's noted that ArmA II servers can be real CPU hogs at times, as well. Are resource spikes that could choke other games common?

4. Is DayZ worth running inside a virtual machine given (2) and (3)? While better safe than sorry is generally our mantra with regards to VMs and new software being given a sandbox period on their own we like to keep our long term choices all inside one instance to make administration as flat and simple as possible.

Thanks for any feedback in advance

Share this post


Link to post
Share on other sites

you will generally need at least 5-6Mbps upstream bandwidth for a 50 player server, if you can't handle sustained amount of high data transfer you may not be able to host a smooth, reliable server.

The ArmA 2 server is actually very stable, you are using beta builds so obviously, as with ANY program, will be more unstable. I have no idea where you got this impression.

ArmA 2 will use a different amount of CPU depending on which CPU you have, mainly it's speed and the way it handles data. More efficient, newer, processor will have lower cpu usage.

The DayZ server I manage is on a virtual machine, well virtualized machine, we use a vmware product which allows low-level access to all the instances of operating systems. There is a free, similar, version of the product we use called vmware sphere. Please note when installing such software you will only have terminal access to the physical machine, you will essentially only be able to remotely access the machine if you want to enter an OS.

Virtual box is generally a decent option for hosting servers, but there are much better solution that offer higher performance and reliability.

Edited by icomrade

Share this post


Link to post
Share on other sites

Thanks for the response.

you will generally need at least 5-6Mbps upstream bandwidth for a 50 player server, if you can't handle sustained amount of high data transfer you may not be able to host a smooth, reliable server.

The university itself has a gigabit line, but we're on a part which is at FE. Actual performance is closer to 90 Mbps down 60 up generally, which is fine there. My real question, I guess, is what is generating that traffic. Is created by players in the server or from stats streamed to and from the server constantly?

The ArmA 2 server is actually very stable, you are using beta builds so obviously, as with ANY program, will be more unstable. I have no idea where you got this impression.

The first people I asked were my friends who work in the game server hosting industry and that's the response they gave right after hearing it was an ArmA II mod. Could have been biased, though, as I know they think that anything not coded natively for Linux or done by either Valve or id inherently isn't coded well in the first place.

ArmA 2 will use a different amount of CPU depending on which CPU you have, mainly it's speed and the way it handles data. More efficient, newer, processor will have lower cpu usage.

We rebuilt our server last Feb with a new 3.2 GHz quad core Sandy Bridge. CPU utilization hasn't been an issue with the new build so far, but then again it wasn't on our older Core 2 quad server either. My basis here was that we had to be careful with, for example, Minecraft in that up until just a few months ago it had memory leaks where the system's entire 16 GB would be taken up.

The DayZ server I manage is on a virtual machine, well virtualized machine, we use a vmware product which allows low-level access to all the instances of operating systems. There is a free, similar, version of the product we use called vmware sphere. Please note when installing such software you will only have terminal access to the physical machine, you will essentially only be able to remotely access the machine if you want to enter an OS.

We use Windows Server's native VM service but, again, we only use it for temporary purposes while testing stability and such. So far we've never kept a program in the VM sandbox for more than a month. Ideally we wouldn't need to even use one.

Virtual box is generally a decent option for hosting servers, but there are much better solution that offer higher performance and reliability.

I'm interested, have an example?

Share this post


Link to post
Share on other sites

Thanks for the response.

The university itself has a gigabit line, but we're on a part which is at FE. Actual performance is closer to 90 Mbps down 60 up generally, which is fine there. My real question, I guess, is what is generating that traffic. Is created by players in the server or from stats streamed to and from the server constantly?

The first people I asked were my friends who work in the game server hosting industry and that's the response they gave right after hearing it was an ArmA II mod. Could have been biased, though, as I know they think that anything not coded natively for Linux or done by either Valve or id inherently isn't coded well in the first place.

We rebuilt our server last Feb with a new 3.2 GHz quad core Sandy Bridge. CPU utilization hasn't been an issue with the new build so far, but then again it wasn't on our older Core 2 quad server either. My basis here was that we had to be careful with, for example, Minecraft in that up until just a few months ago it had memory leaks where the system's entire 16 GB would be taken up.

We use Windows Server's native VM service but, again, we only use it for temporary purposes while testing stability and such. So far we've never kept a program in the VM sandbox for more than a month. Ideally we wouldn't need to even use one.

I'm interested, have an example?

Virtual box is a hypervisor type 2 which means that it has to run ontop of a host operating system, which uses ram and processing power, which creates a small amount of overhead. On the other hand VMware ESXi/VSphere as well as other are type 1 hypervisors which means it does not have any overhead and is able to directly access all resources and components.

Not to mention, having setup both, that virtual box is pretty slow compared to type 1 hypervisors, like VSphere/ESXI; and have a slightly increased latency and decreased network efficiency because, in my set up at least, the network adapter is shared between guest and host OS i.e. one port for the 2 machines.

On the matter of server stability, I have only had 1 or 2 crashes (1 yesterday) in the month the server has been up 24/7. Obviously this is normal running betas, running the stable version the server has never crashed once in over a month. Also, I'm fairly sure that the A2 OA server is native to linux, and compiled on linux as well, at least for the newer versions; IDK completely as I don't host on linux since ArmA 2 OA server is not compatible with DayZ (which may change soon).

As for bandwith I don't know your limit, which would be the only problem, so you should contact your IT or someone that would know if it's limited. The traffic is simply generated due to the way DayZ, and ArmA 2, allow reporting of data back to the server. I'm not sure what exactly the issue is but I know that it has been greatly improved recently, the only thing that can help is more DayZ updates.

CPU utilization shouldn't be a problem, you can lock ArmA 2 to use only a certain number of cores using -exthreads http://community.bistudio.com/wiki/Arma2:_Startup_Parameters#exThreads The default is -exthreads=7 for a quad core cpu, recommended is -exthreads=1 for more than one server. Alternatively you can lock the process to certain cores by setting CPU affinity in task manager.

Edited by icomrade

Share this post


Link to post
Share on other sites

×