In this blog post we'll show you around a bit on our new tactics to improve server security, performances and stability.

To start off there are some hardware changes we have planned on short term.

Our main database/web/etc server MrWhite will receive a new hardware RAID controller. Currently it only has a soft-raid controller on the mobo which is not supported by Linux or VMware. To bypass this we have created a software RAID1+0 in our Linux VM setup. This works all right but its far off the desired performance. Also the windows VM and VMware itself are not in a raid setup this way causing an extra weak point in terms of data security(backups).

For overall security we will be adding a Cisco ASA hardware firewall (5505) to the datacenter on short term. This will provide us with a secure environment were all the important stuff such as SSH, MySQL, SFTP etc. will only be accessible trough a secure VPN tunnel. This makes it possible for us to create a more open environment between the servers itself. With the firewall in place it will also be more easy for us in the future to IP ban abusers on our networks whom might have found exploits to bypass server bans and such. Another feature we're able to implement on the firewall is called traffic shaping. This way we can ensure that for example map downloads won't fully consume the available bandwidth at that time causing other services to slightly lag out.

For the servers overall performances and stability our techtech has rolled out new systems to monitor all possible performance reference data. Build on existing tools such as CACTI and we're now able to map server loads, fps, player counts etc.

Here are some example graphs on what we currently monitor. The goal of this monitoring is to create a more stable server sided FPS as high as possible (target is 66 FPS as minimum as that is what the srcds engine requires).

A little explanation with the graphs we'll show.

First off are graphs per gameserver.

Red: CPU usage reported by the game engine
Orange: CPU usage for that server reported by Linux (top)
Blue: The amount of currently connected users
Green: Frames per second the server is running on (needs to be multiplied by 10)

Up next are graphs per machine, these show the overall CPU, memory and bandwidth usage.

With all this data we can now quickly see what the impact is on the servers after we have applied system changes.

One of the main goals in this project is to create a new Linux kernel for the server to provide minimum CPU load per server on a maximum server FPS.

To be able to do all this in a proper way we require proper systems. Here we come to the next project our techteam has rolled out. We now have a system integrated in our website where we manage all our projects, bug reports, feature requests etc.

As you can see as well in the open critical tasks is our website. Last year we announced that we were hoping for a release of the next website that year, but as you all have noticed, the new website is not there yet. All thought we have made a lot of progress in developing our systems, our design project was still going slow. With the recent techteam addition for graphical design we hope that we now should be able release the new website before the next SpA birthday. (yes, we work by VALVe time as well!)